man.c

   1 /*      $Id: man.c,v 1.65 2010/05/12 16:46:28 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 "libman.h"
  31 #include "libmandoc.h"
  32
  33 const   char *const __man_merrnames[WERRMAX] = {
  34         "invalid character", /* WNPRINT */
  35         "invalid manual section", /* WMSEC */
  36         "invalid date format", /* WDATE */
  37         "scope of prior line violated", /* WLNSCOPE */
  38         "over-zealous prior line scope violation", /* WLNSCOPE2 */
  39         "trailing whitespace", /* WTSPACE */
  40         "unterminated quoted parameter", /* WTQUOTE */
  41         "document has no body", /* WNODATA */
  42         "document has no title/section", /* WNOTITLE */
  43         "invalid escape sequence", /* WESCAPE */
  44         "invalid number format", /* WNUMFMT */
  45         "expected block head arguments", /* WHEADARGS */
  46         "expected block body arguments", /* WBODYARGS */
  47         "expected empty block head", /* WNHEADARGS */
  48         "ill-formed macro", /* WMACROFORM */
  49         "scope open on exit", /* WEXITSCOPE */
  50         "no scope context", /* WNOSCOPE */
  51         "literal context already open", /* WOLITERAL */
  52         "no literal context open", /* WNLITERAL */
  53         "invalid nesting of roff declarations", /* WROFFNEST */
  54         "scope in roff instructions broken", /* WROFFSCOPE */
  55         "document title should be uppercase", /* WTITLECASE */
  56         "deprecated comment style", /* WBADCOMMENT */
  57 };
  58
  59 const   char *const __man_macronames[MAN_MAX] = {
  60         "br",           "TH",           "SH",           "SS",
  61         "TP",           "LP",           "PP",           "P",
  62         "IP",           "HP",           "SM",           "SB",
  63         "BI",           "IB",           "BR",           "RB",
  64         "R",            "B",            "I",            "IR",
  65         "RI",           "na",           "i",            "sp",
  66         "nf",           "fi",           "r",            "RE",
  67         "RS",           "DT",           "UC",           "PD",
  68         "Sp",           "Vb",           "Ve",           "de",
  69         "dei",          "am",           "ami",          "ig",
  70         ".",
  71         };
  72
  73 const   char * const *man_macronames = __man_macronames;
  74
  75 static  struct man_node *man_node_alloc(int, int,
  76                                 enum man_type, enum mant);
  77 static  int              man_node_append(struct man *,
  78                                 struct man_node *);
  79 static  void             man_node_free(struct man_node *);
  80 static  void             man_node_unlink(struct man *,
  81                                 struct man_node *);
  82 static  int              man_ptext(struct man *, int, char *);
  83 static  int              man_pmacro(struct man *, int, char *);
  84 static  void             man_free1(struct man *);
  85 static  void             man_alloc1(struct man *);
  86 static  int              macrowarn(struct man *, int, const char *);
  87
  88
  89 const struct man_node *
  90 man_node(const struct man *m)
  91 {
  92
  93         return(MAN_HALT & m->flags ? NULL : m->first);
  94 }
  95
  96
  97 const struct man_meta *
  98 man_meta(const struct man *m)
  99 {
 100
 101         return(MAN_HALT & m->flags ? NULL : &m->meta);
 102 }
 103
 104
 105 void
 106 man_reset(struct man *man)
 107 {
 108
 109         man_free1(man);
 110         man_alloc1(man);
 111 }
 112
 113
 114 void
 115 man_free(struct man *man)
 116 {
 117
 118         man_free1(man);
 119         free(man);
 120 }
 121
 122
 123 struct man *
 124 man_alloc(void *data, int pflags, const struct man_cb *cb)
 125 {
 126         struct man      *p;
 127
 128         p = mandoc_calloc(1, sizeof(struct man));
 129
 130         if (cb)
 131                 memcpy(&p->cb, cb, sizeof(struct man_cb));
 132
 133         man_hash_init();
 134         p->data = data;
 135         p->pflags = pflags;
 136
 137         man_alloc1(p);
 138         return(p);
 139 }
 140
 141
 142 int
 143 man_endparse(struct man *m)
 144 {
 145
 146         if (MAN_HALT & m->flags)
 147                 return(0);
 148         else if (man_macroend(m))
 149                 return(1);
 150         m->flags |= MAN_HALT;
 151         return(0);
 152 }
 153
 154
 155 int
 156 man_parseln(struct man *m, int ln, char *buf)
 157 {
 158
 159         return('.' == *buf || '\'' == *buf ?
 160                         man_pmacro(m, ln, buf) :
 161                         man_ptext(m, ln, buf));
 162 }
 163
 164
 165 static void
 166 man_free1(struct man *man)
 167 {
 168
 169         if (man->first)
 170                 man_node_delete(man, man->first);
 171         if (man->meta.title)
 172                 free(man->meta.title);
 173         if (man->meta.source)
 174                 free(man->meta.source);
 175         if (man->meta.vol)
 176                 free(man->meta.vol);
 177 }
 178
 179
 180 static void
 181 man_alloc1(struct man *m)
 182 {
 183
 184         memset(&m->meta, 0, sizeof(struct man_meta));
 185         m->flags = 0;
 186         m->last = mandoc_calloc(1, sizeof(struct man_node));
 187         m->first = m->last;
 188         m->last->type = MAN_ROOT;
 189         m->last->tok = MAN_MAX;
 190         m->next = MAN_NEXT_CHILD;
 191 }
 192
 193
 194 static int
 195 man_node_append(struct man *man, struct man_node *p)
 196 {
 197
 198         assert(man->last);
 199         assert(man->first);
 200         assert(MAN_ROOT != p->type);
 201
 202         switch (man->next) {
 203         case (MAN_NEXT_SIBLING):
 204                 man->last->next = p;
 205                 p->prev = man->last;
 206                 p->parent = man->last->parent;
 207                 break;
 208         case (MAN_NEXT_CHILD):
 209                 man->last->child = p;
 210                 p->parent = man->last;
 211                 break;
 212         default:
 213                 abort();
 214                 /* NOTREACHED */
 215         }
 216
 217         assert(p->parent);
 218         p->parent->nchild++;
 219
 220         if ( ! man_valid_pre(man, p))
 221                 return(0);
 222
 223         switch (p->type) {
 224         case (MAN_HEAD):
 225                 assert(MAN_BLOCK == p->parent->type);
 226                 p->parent->head = p;
 227                 break;
 228         case (MAN_BODY):
 229                 assert(MAN_BLOCK == p->parent->type);
 230                 p->parent->body = p;
 231                 break;
 232         default:
 233                 break;
 234         }
 235
 236         man->last = p;
 237
 238         switch (p->type) {
 239         case (MAN_TEXT):
 240                 if ( ! man_valid_post(man))
 241                         return(0);
 242                 if ( ! man_action_post(man))
 243                         return(0);
 244                 break;
 245         default:
 246                 break;
 247         }
 248
 249         return(1);
 250 }
 251
 252
 253 static struct man_node *
 254 man_node_alloc(int line, int pos, enum man_type type, enum mant tok)
 255 {
 256         struct man_node *p;
 257
 258         p = mandoc_calloc(1, sizeof(struct man_node));
 259         p->line = line;
 260         p->pos = pos;
 261         p->type = type;
 262         p->tok = tok;
 263         return(p);
 264 }
 265
 266
 267 int
 268 man_elem_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_ELEM, 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_head_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_HEAD, 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_body_alloc(struct man *m, int line, int pos, enum mant tok)
 295 {
 296         struct man_node *p;
 297
 298         p = man_node_alloc(line, pos, MAN_BODY, tok);
 299         if ( ! man_node_append(m, p))
 300                 return(0);
 301         m->next = MAN_NEXT_CHILD;
 302         return(1);
 303 }
 304
 305
 306 int
 307 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 308 {
 309         struct man_node *p;
 310
 311         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 312         if ( ! man_node_append(m, p))
 313                 return(0);
 314         m->next = MAN_NEXT_CHILD;
 315         return(1);
 316 }
 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(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
 368 static int
 369 man_ptext(struct man *m, int line, char *buf)
 370 {
 371         int              i;
 372
 373         /* Ignore bogus comments. */
 374
 375         if ('\\' == buf[0] && '.' == buf[1] && '\"' == buf[2])
 376                 return(man_pwarn(m, line, 0, WBADCOMMENT));
 377
 378         /* Literal free-form text whitespace is preserved. */
 379
 380         if (MAN_LITERAL & m->flags) {
 381                 if ( ! man_word_alloc(m, line, 0, buf))
 382                         return(0);
 383                 goto descope;
 384         }
 385
 386         /* Pump blank lines directly into the backend. */
 387
 388         for (i = 0; ' ' == buf[i]; i++)
 389                 /* Skip leading whitespace. */ ;
 390
 391         if ('\0' == buf[i]) {
 392                 /* Allocate a blank entry. */
 393                 if ( ! man_word_alloc(m, line, 0, ""))
 394                         return(0);
 395                 goto descope;
 396         }
 397
 398         /*
 399          * Warn if the last un-escaped character is whitespace. Then
 400          * strip away the remaining spaces (tabs stay!).
 401          */
 402
 403         i = (int)strlen(buf);
 404         assert(i);
 405
 406         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 407                 if (i > 1 && '\\' != buf[i - 2])
 408                         if ( ! man_pwarn(m, line, i - 1, WTSPACE))
 409                                 return(0);
 410
 411                 for (--i; i && ' ' == buf[i]; i--)
 412                         /* Spin back to non-space. */ ;
 413
 414                 /* Jump ahead of escaped whitespace. */
 415                 i += '\\' == buf[i] ? 2 : 1;
 416
 417                 buf[i] = '\0';
 418         }
 419
 420         if ( ! man_word_alloc(m, line, 0, buf))
 421                 return(0);
 422
 423         /*
 424          * End-of-sentence check.  If the last character is an unescaped
 425          * EOS character, then flag the node as being the end of a
 426          * sentence.  The front-end will know how to interpret this.
 427          */
 428
 429         assert(i);
 430
 431         switch (buf[i - 1]) {
 432         case ('.'):
 433                 if (i > 1 && '\\' == buf[i - 2])
 434                         break;
 435                 /* FALLTHROUGH */
 436         case ('!'):
 437                 /* FALLTHROUGH */
 438         case ('?'):
 439                 m->last->flags |= MAN_EOS;
 440                 break;
 441         default:
 442                 break;
 443
 444         }
 445
 446 descope:
 447         /*
 448          * Co-ordinate what happens with having a next-line scope open:
 449          * first close out the element scope (if applicable), then close
 450          * out the block scope (also if applicable).
 451          */
 452
 453         if (MAN_ELINE & m->flags) {
 454                 m->flags &= ~MAN_ELINE;
 455                 if ( ! man_unscope(m, m->last->parent, WERRMAX))
 456                         return(0);
 457         }
 458
 459         if ( ! (MAN_BLINE & m->flags))
 460                 return(1);
 461         m->flags &= ~MAN_BLINE;
 462
 463         if ( ! man_unscope(m, m->last->parent, WERRMAX))
 464                 return(0);
 465         return(man_body_alloc(m, line, 0, m->last->tok));
 466 }
 467
 468
 469 static int
 470 macrowarn(struct man *m, int ln, const char *buf)
 471 {
 472         if ( ! (MAN_IGN_MACRO & m->pflags))
 473                 return(man_verr(m, ln, 0, "unknown macro: %s%s",
 474                                 buf, strlen(buf) > 3 ? "..." : ""));
 475         return(man_vwarn(m, ln, 0, "unknown macro: %s%s",
 476                                 buf, strlen(buf) > 3 ? "..." : ""));
 477 }
 478
 479
 480 int
 481 man_pmacro(struct man *m, int ln, char *buf)
 482 {
 483         int              i, j, ppos;
 484         enum mant        tok;
 485         char             mac[5];
 486         struct man_node *n;
 487
 488         /* Comments and empties are quickly ignored. */
 489
 490         if ('\0' == buf[1])
 491                 return(1);
 492
 493         i = 1;
 494
 495         /*
 496          * Skip whitespace between the control character and initial
 497          * text.  "Whitespace" is both spaces and tabs.
 498          */
 499
 500         if (' ' == buf[i] || '\t' == buf[i]) {
 501                 i++;
 502                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 503                         i++;
 504                 if ('\0' == buf[i])
 505                         goto out;
 506         }
 507
 508         ppos = i;
 509
 510         /* Copy the first word into a nil-terminated buffer. */
 511
 512         for (j = 0; j < 4; j++, i++) {
 513                 if ('\0' == (mac[j] = buf[i]))
 514                         break;
 515                 else if (' ' == buf[i])
 516                         break;
 517
 518                 /* Check for invalid characters. */
 519
 520                 if (isgraph((u_char)buf[i]))
 521                         continue;
 522                 return(man_perr(m, ln, i, WNPRINT));
 523         }
 524
 525         mac[j] = '\0';
 526
 527         if (j == 4 || j < 1) {
 528                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 529                         (void)man_perr(m, ln, ppos, WMACROFORM);
 530                         goto err;
 531                 }
 532                 if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
 533                         goto err;
 534                 return(1);
 535         }
 536
 537         if (MAN_MAX == (tok = man_hash_find(mac))) {
 538                 if ( ! macrowarn(m, ln, mac))
 539                         goto err;
 540                 return(1);
 541         }
 542
 543         /* The macro is sane.  Jump to the next word. */
 544
 545         while (buf[i] && ' ' == buf[i])
 546                 i++;
 547
 548         /*
 549          * Trailing whitespace.  Note that tabs are allowed to be passed
 550          * into the parser as "text", so we only warn about spaces here.
 551          */
 552
 553         if ('\0' == buf[i] && ' ' == buf[i - 1])
 554                 if ( ! man_pwarn(m, ln, i - 1, WTSPACE))
 555                         goto err;
 556
 557         /*
 558          * Remove prior ELINE macro, as it's being clobbering by a new
 559          * macro.  Note that NSCOPED macros do not close out ELINE
 560          * macros---they don't print text---so we let those slip by.
 561          * NOTE: we don't allow roff blocks (NOCLOSE) to be embedded
 562          * here because that would stipulate blocks as children of
 563          * elements!
 564          */
 565
 566         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 567                         m->flags & MAN_ELINE) {
 568                 assert(MAN_TEXT != m->last->type);
 569
 570                 /*
 571                  * This occurs in the following construction:
 572                  *   .B
 573                  *   .br
 574                  *   .B
 575                  *   .br
 576                  *   I hate man macros.
 577                  * Flat-out disallow this madness.
 578                  */
 579                 if (MAN_NSCOPED & man_macros[m->last->tok].flags)
 580                         return(man_perr(m, ln, ppos, WLNSCOPE));
 581
 582                 n = m->last;
 583
 584                 assert(n);
 585                 assert(NULL == n->child);
 586                 assert(0 == n->nchild);
 587
 588                 if ( ! man_nwarn(m, n, WLNSCOPE))
 589                         return(0);
 590
 591                 man_node_delete(m, n);
 592                 m->flags &= ~MAN_ELINE;
 593         }
 594
 595         /*
 596          * Save the fact that we're in the next-line for a block.  In
 597          * this way, embedded roff instructions can "remember" state
 598          * when they exit.
 599          */
 600
 601         if (MAN_BLINE & m->flags)
 602                 m->flags |= MAN_BPLINE;
 603
 604         /* Call to handler... */
 605
 606         assert(man_macros[tok].fp);
 607         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 608                 goto err;
 609
 610 out:
 611         /*
 612          * We weren't in a block-line scope when entering the
 613          * above-parsed macro, so return.
 614          */
 615
 616         if ( ! (MAN_BPLINE & m->flags)) {
 617                 m->flags &= ~MAN_ILINE;
 618                 return(1);
 619         }
 620         m->flags &= ~MAN_BPLINE;
 621
 622         /*
 623          * If we're in a block scope, then allow this macro to slip by
 624          * without closing scope around it.
 625          */
 626
 627         if (MAN_ILINE & m->flags) {
 628                 m->flags &= ~MAN_ILINE;
 629                 return(1);
 630         }
 631
 632         /*
 633          * If we've opened a new next-line element scope, then return
 634          * now, as the next line will close out the block scope.
 635          */
 636
 637         if (MAN_ELINE & m->flags)
 638                 return(1);
 639
 640         /* Close out the block scope opened in the prior line.  */
 641
 642         assert(MAN_BLINE & m->flags);
 643         m->flags &= ~MAN_BLINE;
 644
 645         if ( ! man_unscope(m, m->last->parent, WERRMAX))
 646                 return(0);
 647         return(man_body_alloc(m, ln, 0, m->last->tok));
 648
 649 err:    /* Error out. */
 650
 651         m->flags |= MAN_HALT;
 652         return(0);
 653 }
 654
 655
 656 int
 657 man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
 658 {
 659         char             buf[256];
 660         va_list          ap;
 661
 662         if (NULL == man->cb.man_err)
 663                 return(0);
 664
 665         va_start(ap, fmt);
 666         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 667         va_end(ap);
 668         return((*man->cb.man_err)(man->data, ln, pos, buf));
 669 }
 670
 671
 672 int
 673 man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 674 {
 675         char             buf[256];
 676         va_list          ap;
 677
 678         if (NULL == man->cb.man_warn)
 679                 return(0);
 680
 681         va_start(ap, fmt);
 682         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 683         va_end(ap);
 684         return((*man->cb.man_warn)(man->data, ln, pos, buf));
 685 }
 686
 687
 688 int
 689 man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 690 {
 691         const char       *p;
 692
 693         p = __man_merrnames[(int)type];
 694         assert(p);
 695
 696         if (iserr)
 697                 return(man_verr(m, line, pos, p));
 698
 699         return(man_vwarn(m, line, pos, p));
 700 }
 701
 702
 703 /*
 704  * Unlink a node from its context.  If "m" is provided, the last parse
 705  * point will also be adjusted accordingly.
 706  */
 707 static void
 708 man_node_unlink(struct man *m, struct man_node *n)
 709 {
 710
 711         /* Adjust siblings. */
 712
 713         if (n->prev)
 714                 n->prev->next = n->next;
 715         if (n->next)
 716                 n->next->prev = n->prev;
 717
 718         /* Adjust parent. */
 719
 720         if (n->parent) {
 721                 n->parent->nchild--;
 722                 if (n->parent->child == n)
 723                         n->parent->child = n->prev ? n->prev : n->next;
 724         }
 725
 726         /* Adjust parse point, if applicable. */
 727
 728         if (m && m->last == n) {
 729                 /*XXX: this can occur when bailing from validation. */
 730                 /*assert(NULL == n->next);*/
 731                 if (n->prev) {
 732                         m->last = n->prev;
 733                         m->next = MAN_NEXT_SIBLING;
 734                 } else {
 735                         m->last = n->parent;
 736                         m->next = MAN_NEXT_CHILD;
 737                 }
 738         }
 739
 740         if (m && m->first == n)
 741                 m->first = NULL;
 742 }