man.c

   1 /*      $Id: man.c,v 1.97 2011/01/12 10:43:22 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(struct man *, int, int,
  48                                 enum man_type, enum mant);
  49 static  int              man_node_append(struct man *,
  50                                 struct man_node *);
  51 static  int              man_span_alloc(struct man *,
  52                                 const struct tbl_span *);
  53 static  void             man_node_free(struct man_node *);
  54 static  void             man_node_unlink(struct man *,
  55                                 struct man_node *);
  56 static  int              man_ptext(struct man *, int, char *, int);
  57 static  int              man_pmacro(struct man *, int, char *, int);
  58 static  void             man_free1(struct man *);
  59 static  void             man_alloc1(struct man *);
  60 static  int              man_descope(struct man *, int, int);
  61
  62
  63 const struct man_node *
  64 man_node(const struct man *m)
  65 {
  66
  67         assert( ! (MAN_HALT & m->flags));
  68         return(m->first);
  69 }
  70
  71
  72 const struct man_meta *
  73 man_meta(const struct man *m)
  74 {
  75
  76         assert( ! (MAN_HALT & m->flags));
  77         return(&m->meta);
  78 }
  79
  80
  81 void
  82 man_reset(struct man *man)
  83 {
  84
  85         man_free1(man);
  86         man_alloc1(man);
  87 }
  88
  89
  90 void
  91 man_free(struct man *man)
  92 {
  93
  94         man_free1(man);
  95         free(man);
  96 }
  97
  98
  99 struct man *
 100 man_alloc(struct regset *regs, void *data, mandocmsg msg)
 101 {
 102         struct man      *p;
 103
 104         p = mandoc_calloc(1, sizeof(struct man));
 105
 106         man_hash_init();
 107         p->data = data;
 108         p->msg = msg;
 109         p->regs = regs;
 110
 111         man_alloc1(p);
 112         return(p);
 113 }
 114
 115
 116 int
 117 man_endparse(struct man *m)
 118 {
 119
 120         assert( ! (MAN_HALT & m->flags));
 121         if (man_macroend(m))
 122                 return(1);
 123         m->flags |= MAN_HALT;
 124         return(0);
 125 }
 126
 127
 128 int
 129 man_parseln(struct man *m, int ln, char *buf, int offs)
 130 {
 131
 132         m->flags |= MAN_NEWLINE;
 133
 134         assert( ! (MAN_HALT & m->flags));
 135         return(('.' == buf[offs] || '\'' == buf[offs]) ?
 136                         man_pmacro(m, ln, buf, offs) :
 137                         man_ptext(m, ln, buf, offs));
 138 }
 139
 140
 141 static void
 142 man_free1(struct man *man)
 143 {
 144
 145         if (man->first)
 146                 man_node_delete(man, man->first);
 147         if (man->meta.title)
 148                 free(man->meta.title);
 149         if (man->meta.source)
 150                 free(man->meta.source);
 151         if (man->meta.rawdate)
 152                 free(man->meta.rawdate);
 153         if (man->meta.vol)
 154                 free(man->meta.vol);
 155         if (man->meta.msec)
 156                 free(man->meta.msec);
 157 }
 158
 159
 160 static void
 161 man_alloc1(struct man *m)
 162 {
 163
 164         memset(&m->meta, 0, sizeof(struct man_meta));
 165         m->flags = 0;
 166         m->last = mandoc_calloc(1, sizeof(struct man_node));
 167         m->first = m->last;
 168         m->last->type = MAN_ROOT;
 169         m->last->tok = MAN_MAX;
 170         m->next = MAN_NEXT_CHILD;
 171 }
 172
 173
 174 static int
 175 man_node_append(struct man *man, struct man_node *p)
 176 {
 177
 178         assert(man->last);
 179         assert(man->first);
 180         assert(MAN_ROOT != p->type);
 181
 182         switch (man->next) {
 183         case (MAN_NEXT_SIBLING):
 184                 man->last->next = p;
 185                 p->prev = man->last;
 186                 p->parent = man->last->parent;
 187                 break;
 188         case (MAN_NEXT_CHILD):
 189                 man->last->child = p;
 190                 p->parent = man->last;
 191                 break;
 192         default:
 193                 abort();
 194                 /* NOTREACHED */
 195         }
 196
 197         assert(p->parent);
 198         p->parent->nchild++;
 199
 200         if ( ! man_valid_pre(man, p))
 201                 return(0);
 202
 203         switch (p->type) {
 204         case (MAN_HEAD):
 205                 assert(MAN_BLOCK == p->parent->type);
 206                 p->parent->head = p;
 207                 break;
 208         case (MAN_BODY):
 209                 assert(MAN_BLOCK == p->parent->type);
 210                 p->parent->body = p;
 211                 break;
 212         default:
 213                 break;
 214         }
 215
 216         man->last = p;
 217
 218         switch (p->type) {
 219         case (MAN_TBL):
 220                 /* FALLTHROUGH */
 221         case (MAN_TEXT):
 222                 if ( ! man_valid_post(man))
 223                         return(0);
 224                 break;
 225         default:
 226                 break;
 227         }
 228
 229         return(1);
 230 }
 231
 232
 233 static struct man_node *
 234 man_node_alloc(struct man *m, int line, int pos,
 235                 enum man_type type, enum mant tok)
 236 {
 237         struct man_node *p;
 238
 239         p = mandoc_calloc(1, sizeof(struct man_node));
 240         p->line = line;
 241         p->pos = pos;
 242         p->type = type;
 243         p->tok = tok;
 244
 245         if (MAN_NEWLINE & m->flags)
 246                 p->flags |= MAN_LINE;
 247         m->flags &= ~MAN_NEWLINE;
 248         return(p);
 249 }
 250
 251
 252 int
 253 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 254 {
 255         struct man_node *p;
 256
 257         p = man_node_alloc(m, line, pos, MAN_ELEM, tok);
 258         if ( ! man_node_append(m, p))
 259                 return(0);
 260         m->next = MAN_NEXT_CHILD;
 261         return(1);
 262 }
 263
 264
 265 int
 266 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 267 {
 268         struct man_node *p;
 269
 270         p = man_node_alloc(m, line, pos, MAN_HEAD, tok);
 271         if ( ! man_node_append(m, p))
 272                 return(0);
 273         m->next = MAN_NEXT_CHILD;
 274         return(1);
 275 }
 276
 277
 278 int
 279 man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 280 {
 281         struct man_node *p;
 282
 283         p = man_node_alloc(m, line, pos, MAN_BODY, tok);
 284         if ( ! man_node_append(m, p))
 285                 return(0);
 286         m->next = MAN_NEXT_CHILD;
 287         return(1);
 288 }
 289
 290
 291 int
 292 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 293 {
 294         struct man_node *p;
 295
 296         p = man_node_alloc(m, line, pos, MAN_BLOCK, tok);
 297         if ( ! man_node_append(m, p))
 298                 return(0);
 299         m->next = MAN_NEXT_CHILD;
 300         return(1);
 301 }
 302
 303 static int
 304 man_span_alloc(struct man *m, const struct tbl_span *span)
 305 {
 306         struct man_node *n;
 307
 308         /* FIXME: grab from span */
 309         n = man_node_alloc(m, 0, 0, MAN_TBL, MAN_MAX);
 310         n->span = span;
 311
 312         if ( ! man_node_append(m, n))
 313                 return(0);
 314
 315         m->next = MAN_NEXT_SIBLING;
 316         return(1);
 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(m, 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 int
 369 man_addspan(struct man *m, const struct tbl_span *sp)
 370 {
 371
 372         assert( ! (MAN_HALT & m->flags));
 373         if ( ! man_span_alloc(m, sp))
 374                 return(0);
 375         return(man_descope(m, 0, 0));
 376 }
 377
 378 static int
 379 man_descope(struct man *m, int line, int offs)
 380 {
 381         /*
 382          * Co-ordinate what happens with having a next-line scope open:
 383          * first close out the element scope (if applicable), then close
 384          * out the block scope (also if applicable).
 385          */
 386
 387         if (MAN_ELINE & m->flags) {
 388                 m->flags &= ~MAN_ELINE;
 389                 if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 390                         return(0);
 391         }
 392
 393         if ( ! (MAN_BLINE & m->flags))
 394                 return(1);
 395         m->flags &= ~MAN_BLINE;
 396
 397         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 398                 return(0);
 399         return(man_body_alloc(m, line, offs, m->last->tok));
 400 }
 401
 402
 403 static int
 404 man_ptext(struct man *m, int line, char *buf, int offs)
 405 {
 406         int              i;
 407
 408         /* Ignore bogus comments. */
 409
 410         if ('\\' == buf[offs] &&
 411                         '.' == buf[offs + 1] &&
 412                         '"' == buf[offs + 2]) {
 413                 man_pmsg(m, line, offs, MANDOCERR_BADCOMMENT);
 414                 return(1);
 415         }
 416
 417         /* Literal free-form text whitespace is preserved. */
 418
 419         if (MAN_LITERAL & m->flags) {
 420                 if ( ! man_word_alloc(m, line, offs, buf + offs))
 421                         return(0);
 422                 return(man_descope(m, line, offs));
 423         }
 424
 425         /* Pump blank lines directly into the backend. */
 426
 427         for (i = offs; ' ' == buf[i]; i++)
 428                 /* Skip leading whitespace. */ ;
 429
 430         if ('\0' == buf[i]) {
 431                 /* Allocate a blank entry. */
 432                 if ( ! man_word_alloc(m, line, offs, ""))
 433                         return(0);
 434                 return(man_descope(m, line, offs));
 435         }
 436
 437         /*
 438          * Warn if the last un-escaped character is whitespace. Then
 439          * strip away the remaining spaces (tabs stay!).
 440          */
 441
 442         i = (int)strlen(buf);
 443         assert(i);
 444
 445         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 446                 if (i > 1 && '\\' != buf[i - 2])
 447                         man_pmsg(m, line, i - 1, MANDOCERR_EOLNSPACE);
 448
 449                 for (--i; i && ' ' == buf[i]; i--)
 450                         /* Spin back to non-space. */ ;
 451
 452                 /* Jump ahead of escaped whitespace. */
 453                 i += '\\' == buf[i] ? 2 : 1;
 454
 455                 buf[i] = '\0';
 456         }
 457
 458         if ( ! man_word_alloc(m, line, offs, buf + offs))
 459                 return(0);
 460
 461         /*
 462          * End-of-sentence check.  If the last character is an unescaped
 463          * EOS character, then flag the node as being the end of a
 464          * sentence.  The front-end will know how to interpret this.
 465          */
 466
 467         assert(i);
 468         if (mandoc_eos(buf, (size_t)i, 0))
 469                 m->last->flags |= MAN_EOS;
 470
 471         return(man_descope(m, line, offs));
 472 }
 473
 474
 475 static int
 476 man_pmacro(struct man *m, int ln, char *buf, int offs)
 477 {
 478         int              i, j, ppos;
 479         enum mant        tok;
 480         char             mac[5];
 481         struct man_node *n;
 482
 483         /* Comments and empties are quickly ignored. */
 484
 485         offs++;
 486
 487         if ('\0' == buf[offs])
 488                 return(1);
 489
 490         i = offs;
 491
 492         /*
 493          * Skip whitespace between the control character and initial
 494          * text.  "Whitespace" is both spaces and tabs.
 495          */
 496
 497         if (' ' == buf[i] || '\t' == buf[i]) {
 498                 i++;
 499                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 500                         i++;
 501                 if ('\0' == buf[i])
 502                         goto out;
 503         }
 504
 505         ppos = i;
 506
 507         /*
 508          * Copy the first word into a nil-terminated buffer.
 509          * Stop copying when a tab, space, or eoln is encountered.
 510          */
 511
 512         j = 0;
 513         while (j < 4 && '\0' != buf[i] && ' ' != buf[i] && '\t' != buf[i])
 514                 mac[j++] = buf[i++];
 515         mac[j] = '\0';
 516
 517         tok = (j > 0 && j < 4) ? man_hash_find(mac) : MAN_MAX;
 518         if (MAN_MAX == tok) {
 519                 man_vmsg(m, MANDOCERR_MACRO, ln, ppos, "%s", buf + ppos - 1);
 520                 return(1);
 521         }
 522
 523         /* The macro is sane.  Jump to the next word. */
 524
 525         while (buf[i] && ' ' == buf[i])
 526                 i++;
 527
 528         /*
 529          * Trailing whitespace.  Note that tabs are allowed to be passed
 530          * into the parser as "text", so we only warn about spaces here.
 531          */
 532
 533         if ('\0' == buf[i] && ' ' == buf[i - 1])
 534                 man_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE);
 535
 536         /*
 537          * Remove prior ELINE macro, as it's being clobbered by a new
 538          * macro.  Note that NSCOPED macros do not close out ELINE
 539          * macros---they don't print text---so we let those slip by.
 540          */
 541
 542         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 543                         m->flags & MAN_ELINE) {
 544                 n = m->last;
 545                 assert(MAN_TEXT != n->type);
 546
 547                 /* Remove repeated NSCOPED macros causing ELINE. */
 548
 549                 if (MAN_NSCOPED & man_macros[n->tok].flags)
 550                         n = n->parent;
 551
 552                 man_vmsg(m, MANDOCERR_LINESCOPE, n->line, n->pos,
 553                                 "%s", man_macronames[n->tok]);
 554
 555                 man_node_delete(m, n);
 556                 m->flags &= ~MAN_ELINE;
 557         }
 558
 559         /*
 560          * Save the fact that we're in the next-line for a block.  In
 561          * this way, embedded roff instructions can "remember" state
 562          * when they exit.
 563          */
 564
 565         if (MAN_BLINE & m->flags)
 566                 m->flags |= MAN_BPLINE;
 567
 568         /* Call to handler... */
 569
 570         assert(man_macros[tok].fp);
 571         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 572                 goto err;
 573
 574 out:
 575         /*
 576          * We weren't in a block-line scope when entering the
 577          * above-parsed macro, so return.
 578          */
 579
 580         if ( ! (MAN_BPLINE & m->flags)) {
 581                 m->flags &= ~MAN_ILINE;
 582                 return(1);
 583         }
 584         m->flags &= ~MAN_BPLINE;
 585
 586         /*
 587          * If we're in a block scope, then allow this macro to slip by
 588          * without closing scope around it.
 589          */
 590
 591         if (MAN_ILINE & m->flags) {
 592                 m->flags &= ~MAN_ILINE;
 593                 return(1);
 594         }
 595
 596         /*
 597          * If we've opened a new next-line element scope, then return
 598          * now, as the next line will close out the block scope.
 599          */
 600
 601         if (MAN_ELINE & m->flags)
 602                 return(1);
 603
 604         /* Close out the block scope opened in the prior line.  */
 605
 606         assert(MAN_BLINE & m->flags);
 607         m->flags &= ~MAN_BLINE;
 608
 609         if ( ! man_unscope(m, m->last->parent, MANDOCERR_MAX))
 610                 return(0);
 611         return(man_body_alloc(m, ln, offs, m->last->tok));
 612
 613 err:    /* Error out. */
 614
 615         m->flags |= MAN_HALT;
 616         return(0);
 617 }
 618
 619
 620 int
 621 man_vmsg(struct man *man, enum mandocerr t,
 622                 int ln, int pos, const char *fmt, ...)
 623 {
 624         char             buf[256];
 625         va_list          ap;
 626
 627         va_start(ap, fmt);
 628         vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 629         va_end(ap);
 630         return((*man->msg)(t, man->data, ln, pos, buf));
 631 }
 632
 633
 634 /*
 635  * Unlink a node from its context.  If "m" is provided, the last parse
 636  * point will also be adjusted accordingly.
 637  */
 638 static void
 639 man_node_unlink(struct man *m, struct man_node *n)
 640 {
 641
 642         /* Adjust siblings. */
 643
 644         if (n->prev)
 645                 n->prev->next = n->next;
 646         if (n->next)
 647                 n->next->prev = n->prev;
 648
 649         /* Adjust parent. */
 650
 651         if (n->parent) {
 652                 n->parent->nchild--;
 653                 if (n->parent->child == n)
 654                         n->parent->child = n->prev ? n->prev : n->next;
 655         }
 656
 657         /* Adjust parse point, if applicable. */
 658
 659         if (m && m->last == n) {
 660                 /*XXX: this can occur when bailing from validation. */
 661                 /*assert(NULL == n->next);*/
 662                 if (n->prev) {
 663                         m->last = n->prev;
 664                         m->next = MAN_NEXT_SIBLING;
 665                 } else {
 666                         m->last = n->parent;
 667                         m->next = MAN_NEXT_CHILD;
 668                 }
 669         }
 670
 671         if (m && m->first == n)
 672                 m->first = NULL;
 673 }