man.c

   1 /*      $Id: man.c,v 1.69 2010/05/15 20:51:40 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",          ".",
  69         };
  70
  71 const   char * const *man_macronames = __man_macronames;
  72
  73 static  struct man_node *man_node_alloc(int, int,
  74                                 enum man_type, enum mant);
  75 static  int              man_node_append(struct man *,
  76                                 struct man_node *);
  77 static  void             man_node_free(struct man_node *);
  78 static  void             man_node_unlink(struct man *,
  79                                 struct man_node *);
  80 static  int              man_ptext(struct man *, int, char *);
  81 static  int              man_pmacro(struct man *, int, char *);
  82 static  void             man_free1(struct man *);
  83 static  void             man_alloc1(struct man *);
  84 static  int              macrowarn(struct man *, int, const char *);
  85
  86
  87 const struct man_node *
  88 man_node(const struct man *m)
  89 {
  90
  91         return(MAN_HALT & m->flags ? NULL : m->first);
  92 }
  93
  94
  95 const struct man_meta *
  96 man_meta(const struct man *m)
  97 {
  98
  99         return(MAN_HALT & m->flags ? NULL : &m->meta);
 100 }
 101
 102
 103 void
 104 man_reset(struct man *man)
 105 {
 106
 107         man_free1(man);
 108         man_alloc1(man);
 109 }
 110
 111
 112 void
 113 man_free(struct man *man)
 114 {
 115
 116         man_free1(man);
 117         free(man);
 118 }
 119
 120
 121 struct man *
 122 man_alloc(void *data, int pflags, const struct man_cb *cb)
 123 {
 124         struct man      *p;
 125
 126         p = mandoc_calloc(1, sizeof(struct man));
 127
 128         if (cb)
 129                 memcpy(&p->cb, cb, sizeof(struct man_cb));
 130
 131         man_hash_init();
 132         p->data = data;
 133         p->pflags = pflags;
 134
 135         man_alloc1(p);
 136         return(p);
 137 }
 138
 139
 140 int
 141 man_endparse(struct man *m)
 142 {
 143
 144         if (MAN_HALT & m->flags)
 145                 return(0);
 146         else if (man_macroend(m))
 147                 return(1);
 148         m->flags |= MAN_HALT;
 149         return(0);
 150 }
 151
 152
 153 int
 154 man_parseln(struct man *m, int ln, char *buf)
 155 {
 156
 157         return('.' == *buf || '\'' == *buf ?
 158                         man_pmacro(m, ln, buf) :
 159                         man_ptext(m, ln, buf));
 160 }
 161
 162
 163 static void
 164 man_free1(struct man *man)
 165 {
 166
 167         if (man->first)
 168                 man_node_delete(man, man->first);
 169         if (man->meta.title)
 170                 free(man->meta.title);
 171         if (man->meta.source)
 172                 free(man->meta.source);
 173         if (man->meta.vol)
 174                 free(man->meta.vol);
 175         if (man->meta.msec)
 176                 free(man->meta.msec);
 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         /* FIXME: chain of close delims. */
 430
 431         assert(i);
 432
 433         if (mandoc_eos(buf, (size_t)i))
 434                 m->last->flags |= MAN_EOS;
 435
 436 descope:
 437         /*
 438          * Co-ordinate what happens with having a next-line scope open:
 439          * first close out the element scope (if applicable), then close
 440          * out the block scope (also if applicable).
 441          */
 442
 443         if (MAN_ELINE & m->flags) {
 444                 m->flags &= ~MAN_ELINE;
 445                 if ( ! man_unscope(m, m->last->parent, WERRMAX))
 446                         return(0);
 447         }
 448
 449         if ( ! (MAN_BLINE & m->flags))
 450                 return(1);
 451         m->flags &= ~MAN_BLINE;
 452
 453         if ( ! man_unscope(m, m->last->parent, WERRMAX))
 454                 return(0);
 455         return(man_body_alloc(m, line, 0, m->last->tok));
 456 }
 457
 458
 459 static int
 460 macrowarn(struct man *m, int ln, const char *buf)
 461 {
 462         if ( ! (MAN_IGN_MACRO & m->pflags))
 463                 return(man_verr(m, ln, 0, "unknown macro: %s%s",
 464                                 buf, strlen(buf) > 3 ? "..." : ""));
 465         return(man_vwarn(m, ln, 0, "unknown macro: %s%s",
 466                                 buf, strlen(buf) > 3 ? "..." : ""));
 467 }
 468
 469
 470 int
 471 man_pmacro(struct man *m, int ln, char *buf)
 472 {
 473         int              i, j, ppos;
 474         enum mant        tok;
 475         char             mac[5];
 476         struct man_node *n;
 477
 478         /* Comments and empties are quickly ignored. */
 479
 480         if ('\0' == buf[1])
 481                 return(1);
 482
 483         i = 1;
 484
 485         /*
 486          * Skip whitespace between the control character and initial
 487          * text.  "Whitespace" is both spaces and tabs.
 488          */
 489
 490         if (' ' == buf[i] || '\t' == buf[i]) {
 491                 i++;
 492                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 493                         i++;
 494                 if ('\0' == buf[i])
 495                         goto out;
 496         }
 497
 498         ppos = i;
 499
 500         /* Copy the first word into a nil-terminated buffer. */
 501
 502         for (j = 0; j < 4; j++, i++) {
 503                 if ('\0' == (mac[j] = buf[i]))
 504                         break;
 505                 else if (' ' == buf[i])
 506                         break;
 507
 508                 /* Check for invalid characters. */
 509
 510                 if (isgraph((u_char)buf[i]))
 511                         continue;
 512                 return(man_perr(m, ln, i, WNPRINT));
 513         }
 514
 515         mac[j] = '\0';
 516
 517         if (j == 4 || j < 1) {
 518                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 519                         (void)man_perr(m, ln, ppos, WMACROFORM);
 520                         goto err;
 521                 }
 522                 if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
 523                         goto err;
 524                 return(1);
 525         }
 526
 527         if (MAN_MAX == (tok = man_hash_find(mac))) {
 528                 if ( ! macrowarn(m, ln, mac))
 529                         goto err;
 530                 return(1);
 531         }
 532
 533         /* The macro is sane.  Jump to the next word. */
 534
 535         while (buf[i] && ' ' == buf[i])
 536                 i++;
 537
 538         /*
 539          * Trailing whitespace.  Note that tabs are allowed to be passed
 540          * into the parser as "text", so we only warn about spaces here.
 541          */
 542
 543         if ('\0' == buf[i] && ' ' == buf[i - 1])
 544                 if ( ! man_pwarn(m, ln, i - 1, WTSPACE))
 545                         goto err;
 546
 547         /*
 548          * Remove prior ELINE macro, as it's being clobbering by a new
 549          * macro.  Note that NSCOPED macros do not close out ELINE
 550          * macros---they don't print text---so we let those slip by.
 551          * NOTE: we don't allow roff blocks (NOCLOSE) to be embedded
 552          * here because that would stipulate blocks as children of
 553          * elements!
 554          */
 555
 556         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 557                         m->flags & MAN_ELINE) {
 558                 assert(MAN_TEXT != m->last->type);
 559
 560                 /*
 561                  * This occurs in the following construction:
 562                  *   .B
 563                  *   .br
 564                  *   .B
 565                  *   .br
 566                  *   I hate man macros.
 567                  * Flat-out disallow this madness.
 568                  */
 569                 if (MAN_NSCOPED & man_macros[m->last->tok].flags)
 570                         return(man_perr(m, ln, ppos, WLNSCOPE));
 571
 572                 n = m->last;
 573
 574                 assert(n);
 575                 assert(NULL == n->child);
 576                 assert(0 == n->nchild);
 577
 578                 if ( ! man_nwarn(m, n, WLNSCOPE))
 579                         return(0);
 580
 581                 man_node_delete(m, n);
 582                 m->flags &= ~MAN_ELINE;
 583         }
 584
 585         /*
 586          * Save the fact that we're in the next-line for a block.  In
 587          * this way, embedded roff instructions can "remember" state
 588          * when they exit.
 589          */
 590
 591         if (MAN_BLINE & m->flags)
 592                 m->flags |= MAN_BPLINE;
 593
 594         /* Call to handler... */
 595
 596         assert(man_macros[tok].fp);
 597         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 598                 goto err;
 599
 600 out:
 601         /*
 602          * We weren't in a block-line scope when entering the
 603          * above-parsed macro, so return.
 604          */
 605
 606         if ( ! (MAN_BPLINE & m->flags)) {
 607                 m->flags &= ~MAN_ILINE;
 608                 return(1);
 609         }
 610         m->flags &= ~MAN_BPLINE;
 611
 612         /*
 613          * If we're in a block scope, then allow this macro to slip by
 614          * without closing scope around it.
 615          */
 616
 617         if (MAN_ILINE & m->flags) {
 618                 m->flags &= ~MAN_ILINE;
 619                 return(1);
 620         }
 621
 622         /*
 623          * If we've opened a new next-line element scope, then return
 624          * now, as the next line will close out the block scope.
 625          */
 626
 627         if (MAN_ELINE & m->flags)
 628                 return(1);
 629
 630         /* Close out the block scope opened in the prior line.  */
 631
 632         assert(MAN_BLINE & m->flags);
 633         m->flags &= ~MAN_BLINE;
 634
 635         if ( ! man_unscope(m, m->last->parent, WERRMAX))
 636                 return(0);
 637         return(man_body_alloc(m, ln, 0, m->last->tok));
 638
 639 err:    /* Error out. */
 640
 641         m->flags |= MAN_HALT;
 642         return(0);
 643 }
 644
 645
 646 int
 647 man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
 648 {
 649         char             buf[256];
 650         va_list          ap;
 651
 652         if (NULL == man->cb.man_err)
 653                 return(0);
 654
 655         va_start(ap, fmt);
 656         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 657         va_end(ap);
 658         return((*man->cb.man_err)(man->data, ln, pos, buf));
 659 }
 660
 661
 662 int
 663 man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 664 {
 665         char             buf[256];
 666         va_list          ap;
 667
 668         if (NULL == man->cb.man_warn)
 669                 return(0);
 670
 671         va_start(ap, fmt);
 672         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 673         va_end(ap);
 674         return((*man->cb.man_warn)(man->data, ln, pos, buf));
 675 }
 676
 677
 678 int
 679 man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 680 {
 681         const char       *p;
 682
 683         p = __man_merrnames[(int)type];
 684         assert(p);
 685
 686         if (iserr)
 687                 return(man_verr(m, line, pos, p));
 688
 689         return(man_vwarn(m, line, pos, p));
 690 }
 691
 692
 693 /*
 694  * Unlink a node from its context.  If "m" is provided, the last parse
 695  * point will also be adjusted accordingly.
 696  */
 697 static void
 698 man_node_unlink(struct man *m, struct man_node *n)
 699 {
 700
 701         /* Adjust siblings. */
 702
 703         if (n->prev)
 704                 n->prev->next = n->next;
 705         if (n->next)
 706                 n->next->prev = n->prev;
 707
 708         /* Adjust parent. */
 709
 710         if (n->parent) {
 711                 n->parent->nchild--;
 712                 if (n->parent->child == n)
 713                         n->parent->child = n->prev ? n->prev : n->next;
 714         }
 715
 716         /* Adjust parse point, if applicable. */
 717
 718         if (m && m->last == n) {
 719                 /*XXX: this can occur when bailing from validation. */
 720                 /*assert(NULL == n->next);*/
 721                 if (n->prev) {
 722                         m->last = n->prev;
 723                         m->next = MAN_NEXT_SIBLING;
 724                 } else {
 725                         m->last = n->parent;
 726                         m->next = MAN_NEXT_CHILD;
 727                 }
 728         }
 729
 730         if (m && m->first == n)
 731                 m->first = NULL;
 732 }