man.c

   1 /*      $Id: man.c,v 1.68 2010/05/15 15:54:39 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 date format", /* WDATE */
  36         "scope of prior line violated", /* WLNSCOPE */
  37         "over-zealous prior line scope violation", /* WLNSCOPE2 */
  38         "trailing whitespace", /* WTSPACE */
  39         "unterminated quoted parameter", /* WTQUOTE */
  40         "document has no body", /* WNODATA */
  41         "document has no title/section", /* WNOTITLE */
  42         "invalid escape sequence", /* WESCAPE */
  43         "invalid number format", /* WNUMFMT */
  44         "expected block head arguments", /* WHEADARGS */
  45         "expected block body arguments", /* WBODYARGS */
  46         "expected empty block head", /* WNHEADARGS */
  47         "ill-formed macro", /* WMACROFORM */
  48         "scope open on exit", /* WEXITSCOPE */
  49         "no scope context", /* WNOSCOPE */
  50         "literal context already open", /* WOLITERAL */
  51         "no literal context open", /* WNLITERAL */
  52         "invalid nesting of roff declarations", /* WROFFNEST */
  53         "scope in roff instructions broken", /* WROFFSCOPE */
  54         "document title should be uppercase", /* WTITLECASE */
  55         "deprecated comment style", /* WBADCOMMENT */
  56 };
  57
  58 const   char *const __man_macronames[MAN_MAX] = {
  59         "br",           "TH",           "SH",           "SS",
  60         "TP",           "LP",           "PP",           "P",
  61         "IP",           "HP",           "SM",           "SB",
  62         "BI",           "IB",           "BR",           "RB",
  63         "R",            "B",            "I",            "IR",
  64         "RI",           "na",           "i",            "sp",
  65         "nf",           "fi",           "r",            "RE",
  66         "RS",           "DT",           "UC",           "PD",
  67         "Sp",           "Vb",           "Ve",           "de",
  68         "dei",          "am",           "ami",          "ig",
  69         ".",
  70         };
  71
  72 const   char * const *man_macronames = __man_macronames;
  73
  74 static  struct man_node *man_node_alloc(int, int,
  75                                 enum man_type, enum mant);
  76 static  int              man_node_append(struct man *,
  77                                 struct man_node *);
  78 static  void             man_node_free(struct man_node *);
  79 static  void             man_node_unlink(struct man *,
  80                                 struct man_node *);
  81 static  int              man_ptext(struct man *, int, char *);
  82 static  int              man_pmacro(struct man *, int, char *);
  83 static  void             man_free1(struct man *);
  84 static  void             man_alloc1(struct man *);
  85 static  int              macrowarn(struct man *, int, const char *);
  86
  87
  88 const struct man_node *
  89 man_node(const struct man *m)
  90 {
  91
  92         return(MAN_HALT & m->flags ? NULL : m->first);
  93 }
  94
  95
  96 const struct man_meta *
  97 man_meta(const struct man *m)
  98 {
  99
 100         return(MAN_HALT & m->flags ? NULL : &m->meta);
 101 }
 102
 103
 104 void
 105 man_reset(struct man *man)
 106 {
 107
 108         man_free1(man);
 109         man_alloc1(man);
 110 }
 111
 112
 113 void
 114 man_free(struct man *man)
 115 {
 116
 117         man_free1(man);
 118         free(man);
 119 }
 120
 121
 122 struct man *
 123 man_alloc(void *data, int pflags, const struct man_cb *cb)
 124 {
 125         struct man      *p;
 126
 127         p = mandoc_calloc(1, sizeof(struct man));
 128
 129         if (cb)
 130                 memcpy(&p->cb, cb, sizeof(struct man_cb));
 131
 132         man_hash_init();
 133         p->data = data;
 134         p->pflags = pflags;
 135
 136         man_alloc1(p);
 137         return(p);
 138 }
 139
 140
 141 int
 142 man_endparse(struct man *m)
 143 {
 144
 145         if (MAN_HALT & m->flags)
 146                 return(0);
 147         else if (man_macroend(m))
 148                 return(1);
 149         m->flags |= MAN_HALT;
 150         return(0);
 151 }
 152
 153
 154 int
 155 man_parseln(struct man *m, int ln, char *buf)
 156 {
 157
 158         return('.' == *buf || '\'' == *buf ?
 159                         man_pmacro(m, ln, buf) :
 160                         man_ptext(m, ln, buf));
 161 }
 162
 163
 164 static void
 165 man_free1(struct man *man)
 166 {
 167
 168         if (man->first)
 169                 man_node_delete(man, man->first);
 170         if (man->meta.title)
 171                 free(man->meta.title);
 172         if (man->meta.source)
 173                 free(man->meta.source);
 174         if (man->meta.vol)
 175                 free(man->meta.vol);
 176         if (man->meta.msec)
 177                 free(man->meta.msec);
 178 }
 179
 180
 181 static void
 182 man_alloc1(struct man *m)
 183 {
 184
 185         memset(&m->meta, 0, sizeof(struct man_meta));
 186         m->flags = 0;
 187         m->last = mandoc_calloc(1, sizeof(struct man_node));
 188         m->first = m->last;
 189         m->last->type = MAN_ROOT;
 190         m->last->tok = MAN_MAX;
 191         m->next = MAN_NEXT_CHILD;
 192 }
 193
 194
 195 static int
 196 man_node_append(struct man *man, struct man_node *p)
 197 {
 198
 199         assert(man->last);
 200         assert(man->first);
 201         assert(MAN_ROOT != p->type);
 202
 203         switch (man->next) {
 204         case (MAN_NEXT_SIBLING):
 205                 man->last->next = p;
 206                 p->prev = man->last;
 207                 p->parent = man->last->parent;
 208                 break;
 209         case (MAN_NEXT_CHILD):
 210                 man->last->child = p;
 211                 p->parent = man->last;
 212                 break;
 213         default:
 214                 abort();
 215                 /* NOTREACHED */
 216         }
 217
 218         assert(p->parent);
 219         p->parent->nchild++;
 220
 221         if ( ! man_valid_pre(man, p))
 222                 return(0);
 223
 224         switch (p->type) {
 225         case (MAN_HEAD):
 226                 assert(MAN_BLOCK == p->parent->type);
 227                 p->parent->head = p;
 228                 break;
 229         case (MAN_BODY):
 230                 assert(MAN_BLOCK == p->parent->type);
 231                 p->parent->body = p;
 232                 break;
 233         default:
 234                 break;
 235         }
 236
 237         man->last = p;
 238
 239         switch (p->type) {
 240         case (MAN_TEXT):
 241                 if ( ! man_valid_post(man))
 242                         return(0);
 243                 if ( ! man_action_post(man))
 244                         return(0);
 245                 break;
 246         default:
 247                 break;
 248         }
 249
 250         return(1);
 251 }
 252
 253
 254 static struct man_node *
 255 man_node_alloc(int line, int pos, enum man_type type, enum mant tok)
 256 {
 257         struct man_node *p;
 258
 259         p = mandoc_calloc(1, sizeof(struct man_node));
 260         p->line = line;
 261         p->pos = pos;
 262         p->type = type;
 263         p->tok = tok;
 264         return(p);
 265 }
 266
 267
 268 int
 269 man_elem_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_ELEM, 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_head_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_HEAD, 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_body_alloc(struct man *m, int line, int pos, enum mant tok)
 296 {
 297         struct man_node *p;
 298
 299         p = man_node_alloc(line, pos, MAN_BODY, tok);
 300         if ( ! man_node_append(m, p))
 301                 return(0);
 302         m->next = MAN_NEXT_CHILD;
 303         return(1);
 304 }
 305
 306
 307 int
 308 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 309 {
 310         struct man_node *p;
 311
 312         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 313         if ( ! man_node_append(m, p))
 314                 return(0);
 315         m->next = MAN_NEXT_CHILD;
 316         return(1);
 317 }
 318
 319
 320 int
 321 man_word_alloc(struct man *m, int line, int pos, const char *word)
 322 {
 323         struct man_node *n;
 324         size_t           sv, len;
 325
 326         len = strlen(word);
 327
 328         n = man_node_alloc(line, pos, MAN_TEXT, MAN_MAX);
 329         n->string = mandoc_malloc(len + 1);
 330         sv = strlcpy(n->string, word, len + 1);
 331
 332         /* Prohibit truncation. */
 333         assert(sv < len + 1);
 334
 335         if ( ! man_node_append(m, n))
 336                 return(0);
 337
 338         m->next = MAN_NEXT_SIBLING;
 339         return(1);
 340 }
 341
 342
 343 /*
 344  * Free all of the resources held by a node.  This does NOT unlink a
 345  * node from its context; for that, see man_node_unlink().
 346  */
 347 static void
 348 man_node_free(struct man_node *p)
 349 {
 350
 351         if (p->string)
 352                 free(p->string);
 353         free(p);
 354 }
 355
 356
 357 void
 358 man_node_delete(struct man *m, struct man_node *p)
 359 {
 360
 361         while (p->child)
 362                 man_node_delete(m, p->child);
 363
 364         man_node_unlink(m, p);
 365         man_node_free(p);
 366 }
 367
 368
 369 static int
 370 man_ptext(struct man *m, int line, char *buf)
 371 {
 372         int              i;
 373
 374         /* Ignore bogus comments. */
 375
 376         if ('\\' == buf[0] && '.' == buf[1] && '\"' == buf[2])
 377                 return(man_pwarn(m, line, 0, WBADCOMMENT));
 378
 379         /* Literal free-form text whitespace is preserved. */
 380
 381         if (MAN_LITERAL & m->flags) {
 382                 if ( ! man_word_alloc(m, line, 0, buf))
 383                         return(0);
 384                 goto descope;
 385         }
 386
 387         /* Pump blank lines directly into the backend. */
 388
 389         for (i = 0; ' ' == buf[i]; i++)
 390                 /* Skip leading whitespace. */ ;
 391
 392         if ('\0' == buf[i]) {
 393                 /* Allocate a blank entry. */
 394                 if ( ! man_word_alloc(m, line, 0, ""))
 395                         return(0);
 396                 goto descope;
 397         }
 398
 399         /*
 400          * Warn if the last un-escaped character is whitespace. Then
 401          * strip away the remaining spaces (tabs stay!).
 402          */
 403
 404         i = (int)strlen(buf);
 405         assert(i);
 406
 407         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 408                 if (i > 1 && '\\' != buf[i - 2])
 409                         if ( ! man_pwarn(m, line, i - 1, WTSPACE))
 410                                 return(0);
 411
 412                 for (--i; i && ' ' == buf[i]; i--)
 413                         /* Spin back to non-space. */ ;
 414
 415                 /* Jump ahead of escaped whitespace. */
 416                 i += '\\' == buf[i] ? 2 : 1;
 417
 418                 buf[i] = '\0';
 419         }
 420
 421         if ( ! man_word_alloc(m, line, 0, buf))
 422                 return(0);
 423
 424         /*
 425          * End-of-sentence check.  If the last character is an unescaped
 426          * EOS character, then flag the node as being the end of a
 427          * sentence.  The front-end will know how to interpret this.
 428          */
 429
 430         /* FIXME: chain of close delims. */
 431
 432         assert(i);
 433
 434         if (mandoc_eos(buf, (size_t)i))
 435                 m->last->flags |= MAN_EOS;
 436
 437 descope:
 438         /*
 439          * Co-ordinate what happens with having a next-line scope open:
 440          * first close out the element scope (if applicable), then close
 441          * out the block scope (also if applicable).
 442          */
 443
 444         if (MAN_ELINE & m->flags) {
 445                 m->flags &= ~MAN_ELINE;
 446                 if ( ! man_unscope(m, m->last->parent, WERRMAX))
 447                         return(0);
 448         }
 449
 450         if ( ! (MAN_BLINE & m->flags))
 451                 return(1);
 452         m->flags &= ~MAN_BLINE;
 453
 454         if ( ! man_unscope(m, m->last->parent, WERRMAX))
 455                 return(0);
 456         return(man_body_alloc(m, line, 0, m->last->tok));
 457 }
 458
 459
 460 static int
 461 macrowarn(struct man *m, int ln, const char *buf)
 462 {
 463         if ( ! (MAN_IGN_MACRO & m->pflags))
 464                 return(man_verr(m, ln, 0, "unknown macro: %s%s",
 465                                 buf, strlen(buf) > 3 ? "..." : ""));
 466         return(man_vwarn(m, ln, 0, "unknown macro: %s%s",
 467                                 buf, strlen(buf) > 3 ? "..." : ""));
 468 }
 469
 470
 471 int
 472 man_pmacro(struct man *m, int ln, char *buf)
 473 {
 474         int              i, j, ppos;
 475         enum mant        tok;
 476         char             mac[5];
 477         struct man_node *n;
 478
 479         /* Comments and empties are quickly ignored. */
 480
 481         if ('\0' == buf[1])
 482                 return(1);
 483
 484         i = 1;
 485
 486         /*
 487          * Skip whitespace between the control character and initial
 488          * text.  "Whitespace" is both spaces and tabs.
 489          */
 490
 491         if (' ' == buf[i] || '\t' == buf[i]) {
 492                 i++;
 493                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 494                         i++;
 495                 if ('\0' == buf[i])
 496                         goto out;
 497         }
 498
 499         ppos = i;
 500
 501         /* Copy the first word into a nil-terminated buffer. */
 502
 503         for (j = 0; j < 4; j++, i++) {
 504                 if ('\0' == (mac[j] = buf[i]))
 505                         break;
 506                 else if (' ' == buf[i])
 507                         break;
 508
 509                 /* Check for invalid characters. */
 510
 511                 if (isgraph((u_char)buf[i]))
 512                         continue;
 513                 return(man_perr(m, ln, i, WNPRINT));
 514         }
 515
 516         mac[j] = '\0';
 517
 518         if (j == 4 || j < 1) {
 519                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 520                         (void)man_perr(m, ln, ppos, WMACROFORM);
 521                         goto err;
 522                 }
 523                 if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
 524                         goto err;
 525                 return(1);
 526         }
 527
 528         if (MAN_MAX == (tok = man_hash_find(mac))) {
 529                 if ( ! macrowarn(m, ln, mac))
 530                         goto err;
 531                 return(1);
 532         }
 533
 534         /* The macro is sane.  Jump to the next word. */
 535
 536         while (buf[i] && ' ' == buf[i])
 537                 i++;
 538
 539         /*
 540          * Trailing whitespace.  Note that tabs are allowed to be passed
 541          * into the parser as "text", so we only warn about spaces here.
 542          */
 543
 544         if ('\0' == buf[i] && ' ' == buf[i - 1])
 545                 if ( ! man_pwarn(m, ln, i - 1, WTSPACE))
 546                         goto err;
 547
 548         /*
 549          * Remove prior ELINE macro, as it's being clobbering by a new
 550          * macro.  Note that NSCOPED macros do not close out ELINE
 551          * macros---they don't print text---so we let those slip by.
 552          * NOTE: we don't allow roff blocks (NOCLOSE) to be embedded
 553          * here because that would stipulate blocks as children of
 554          * elements!
 555          */
 556
 557         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 558                         m->flags & MAN_ELINE) {
 559                 assert(MAN_TEXT != m->last->type);
 560
 561                 /*
 562                  * This occurs in the following construction:
 563                  *   .B
 564                  *   .br
 565                  *   .B
 566                  *   .br
 567                  *   I hate man macros.
 568                  * Flat-out disallow this madness.
 569                  */
 570                 if (MAN_NSCOPED & man_macros[m->last->tok].flags)
 571                         return(man_perr(m, ln, ppos, WLNSCOPE));
 572
 573                 n = m->last;
 574
 575                 assert(n);
 576                 assert(NULL == n->child);
 577                 assert(0 == n->nchild);
 578
 579                 if ( ! man_nwarn(m, n, WLNSCOPE))
 580                         return(0);
 581
 582                 man_node_delete(m, n);
 583                 m->flags &= ~MAN_ELINE;
 584         }
 585
 586         /*
 587          * Save the fact that we're in the next-line for a block.  In
 588          * this way, embedded roff instructions can "remember" state
 589          * when they exit.
 590          */
 591
 592         if (MAN_BLINE & m->flags)
 593                 m->flags |= MAN_BPLINE;
 594
 595         /* Call to handler... */
 596
 597         assert(man_macros[tok].fp);
 598         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 599                 goto err;
 600
 601 out:
 602         /*
 603          * We weren't in a block-line scope when entering the
 604          * above-parsed macro, so return.
 605          */
 606
 607         if ( ! (MAN_BPLINE & m->flags)) {
 608                 m->flags &= ~MAN_ILINE;
 609                 return(1);
 610         }
 611         m->flags &= ~MAN_BPLINE;
 612
 613         /*
 614          * If we're in a block scope, then allow this macro to slip by
 615          * without closing scope around it.
 616          */
 617
 618         if (MAN_ILINE & m->flags) {
 619                 m->flags &= ~MAN_ILINE;
 620                 return(1);
 621         }
 622
 623         /*
 624          * If we've opened a new next-line element scope, then return
 625          * now, as the next line will close out the block scope.
 626          */
 627
 628         if (MAN_ELINE & m->flags)
 629                 return(1);
 630
 631         /* Close out the block scope opened in the prior line.  */
 632
 633         assert(MAN_BLINE & m->flags);
 634         m->flags &= ~MAN_BLINE;
 635
 636         if ( ! man_unscope(m, m->last->parent, WERRMAX))
 637                 return(0);
 638         return(man_body_alloc(m, ln, 0, m->last->tok));
 639
 640 err:    /* Error out. */
 641
 642         m->flags |= MAN_HALT;
 643         return(0);
 644 }
 645
 646
 647 int
 648 man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
 649 {
 650         char             buf[256];
 651         va_list          ap;
 652
 653         if (NULL == man->cb.man_err)
 654                 return(0);
 655
 656         va_start(ap, fmt);
 657         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 658         va_end(ap);
 659         return((*man->cb.man_err)(man->data, ln, pos, buf));
 660 }
 661
 662
 663 int
 664 man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 665 {
 666         char             buf[256];
 667         va_list          ap;
 668
 669         if (NULL == man->cb.man_warn)
 670                 return(0);
 671
 672         va_start(ap, fmt);
 673         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 674         va_end(ap);
 675         return((*man->cb.man_warn)(man->data, ln, pos, buf));
 676 }
 677
 678
 679 int
 680 man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 681 {
 682         const char       *p;
 683
 684         p = __man_merrnames[(int)type];
 685         assert(p);
 686
 687         if (iserr)
 688                 return(man_verr(m, line, pos, p));
 689
 690         return(man_vwarn(m, line, pos, p));
 691 }
 692
 693
 694 /*
 695  * Unlink a node from its context.  If "m" is provided, the last parse
 696  * point will also be adjusted accordingly.
 697  */
 698 static void
 699 man_node_unlink(struct man *m, struct man_node *n)
 700 {
 701
 702         /* Adjust siblings. */
 703
 704         if (n->prev)
 705                 n->prev->next = n->next;
 706         if (n->next)
 707                 n->next->prev = n->prev;
 708
 709         /* Adjust parent. */
 710
 711         if (n->parent) {
 712                 n->parent->nchild--;
 713                 if (n->parent->child == n)
 714                         n->parent->child = n->prev ? n->prev : n->next;
 715         }
 716
 717         /* Adjust parse point, if applicable. */
 718
 719         if (m && m->last == n) {
 720                 /*XXX: this can occur when bailing from validation. */
 721                 /*assert(NULL == n->next);*/
 722                 if (n->prev) {
 723                         m->last = n->prev;
 724                         m->next = MAN_NEXT_SIBLING;
 725                 } else {
 726                         m->last = n->parent;
 727                         m->next = MAN_NEXT_CHILD;
 728                 }
 729         }
 730
 731         if (m && m->first == n)
 732                 m->first = NULL;
 733 }