man.c

   1 /*      $Id: man.c,v 1.146 2014/12/28 14:42:27 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
   5  * Copyright (c) 2011 Joerg Sonnenberger <joerg@netbsd.org>
   6  *
   7  * Permission to use, copy, modify, and distribute this software for any
   8  * purpose with or without fee is hereby granted, provided that the above
   9  * copyright notice and this permission notice appear in all copies.
  10  *
  11  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  13  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  14  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  15  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  16  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  17  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  18  */
  19 #include "config.h"
  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 "man.h"
  31 #include "mandoc.h"
  32 #include "mandoc_aux.h"
  33 #include "libman.h"
  34 #include "libmandoc.h"
  35
  36 const   char *const __man_macronames[MAN_MAX] = {
  37         "br",           "TH",           "SH",           "SS",
  38         "TP",           "LP",           "PP",           "P",
  39         "IP",           "HP",           "SM",           "SB",
  40         "BI",           "IB",           "BR",           "RB",
  41         "R",            "B",            "I",            "IR",
  42         "RI",           "na",           "sp",           "nf",
  43         "fi",           "RE",           "RS",           "DT",
  44         "UC",           "PD",           "AT",           "in",
  45         "ft",           "OP",           "EX",           "EE",
  46         "UR",           "UE",           "ll"
  47         };
  48
  49 const   char * const *man_macronames = __man_macronames;
  50
  51 static  struct man_node *man_node_alloc(struct man *, int, int,
  52                                 enum man_type, enum mant);
  53 static  void             man_node_append(struct man *, struct man_node *);
  54 static  void             man_node_free(struct man_node *);
  55 static  void             man_node_unlink(struct man *,
  56                                 struct man_node *);
  57 static  int              man_ptext(struct man *, int, char *, int);
  58 static  int              man_pmacro(struct man *, int, char *, int);
  59 static  void             man_free1(struct man *);
  60 static  void             man_alloc1(struct man *);
  61 static  void             man_descope(struct man *, int, int);
  62
  63
  64 const struct man_node *
  65 man_node(const struct man *man)
  66 {
  67
  68         return(man->first);
  69 }
  70
  71 const struct man_meta *
  72 man_meta(const struct man *man)
  73 {
  74
  75         return(&man->meta);
  76 }
  77
  78 void
  79 man_reset(struct man *man)
  80 {
  81
  82         man_free1(man);
  83         man_alloc1(man);
  84 }
  85
  86 void
  87 man_free(struct man *man)
  88 {
  89
  90         man_free1(man);
  91         free(man);
  92 }
  93
  94 struct man *
  95 man_alloc(struct roff *roff, struct mparse *parse,
  96         const char *defos, int quick)
  97 {
  98         struct man      *p;
  99
 100         p = mandoc_calloc(1, sizeof(struct man));
 101
 102         man_hash_init();
 103         p->parse = parse;
 104         p->defos = defos;
 105         p->quick = quick;
 106         p->roff = roff;
 107
 108         man_alloc1(p);
 109         return(p);
 110 }
 111
 112 int
 113 man_endparse(struct man *man)
 114 {
 115
 116         man_macroend(man);
 117         return(1);
 118 }
 119
 120 int
 121 man_parseln(struct man *man, int ln, char *buf, int offs)
 122 {
 123
 124         if (man->last->type != MAN_EQN || ln > man->last->line)
 125                 man->flags |= MAN_NEWLINE;
 126
 127         return (roff_getcontrol(man->roff, buf, &offs) ?
 128             man_pmacro(man, ln, buf, offs) :
 129             man_ptext(man, ln, buf, offs));
 130 }
 131
 132 static void
 133 man_free1(struct man *man)
 134 {
 135
 136         if (man->first)
 137                 man_node_delete(man, man->first);
 138         free(man->meta.title);
 139         free(man->meta.source);
 140         free(man->meta.date);
 141         free(man->meta.vol);
 142         free(man->meta.msec);
 143 }
 144
 145 static void
 146 man_alloc1(struct man *man)
 147 {
 148
 149         memset(&man->meta, 0, sizeof(struct man_meta));
 150         man->flags = 0;
 151         man->last = mandoc_calloc(1, sizeof(struct man_node));
 152         man->first = man->last;
 153         man->last->type = MAN_ROOT;
 154         man->last->tok = MAN_MAX;
 155         man->next = MAN_NEXT_CHILD;
 156 }
 157
 158
 159 static void
 160 man_node_append(struct man *man, struct man_node *p)
 161 {
 162
 163         assert(man->last);
 164         assert(man->first);
 165         assert(p->type != MAN_ROOT);
 166
 167         switch (man->next) {
 168         case MAN_NEXT_SIBLING:
 169                 man->last->next = p;
 170                 p->prev = man->last;
 171                 p->parent = man->last->parent;
 172                 break;
 173         case MAN_NEXT_CHILD:
 174                 man->last->child = p;
 175                 p->parent = man->last;
 176                 break;
 177         default:
 178                 abort();
 179                 /* NOTREACHED */
 180         }
 181
 182         assert(p->parent);
 183         p->parent->nchild++;
 184
 185         switch (p->type) {
 186         case MAN_BLOCK:
 187                 if (p->tok == MAN_SH || p->tok == MAN_SS)
 188                         man->flags &= ~MAN_LITERAL;
 189                 break;
 190         case MAN_HEAD:
 191                 assert(p->parent->type == MAN_BLOCK);
 192                 p->parent->head = p;
 193                 break;
 194         case MAN_BODY:
 195                 assert(p->parent->type == MAN_BLOCK);
 196                 p->parent->body = p;
 197                 break;
 198         default:
 199                 break;
 200         }
 201
 202         man->last = p;
 203
 204         switch (p->type) {
 205         case MAN_TBL:
 206                 /* FALLTHROUGH */
 207         case MAN_TEXT:
 208                 man_valid_post(man);
 209                 break;
 210         default:
 211                 break;
 212         }
 213 }
 214
 215 static struct man_node *
 216 man_node_alloc(struct man *man, int line, int pos,
 217                 enum man_type type, enum mant tok)
 218 {
 219         struct man_node *p;
 220
 221         p = mandoc_calloc(1, sizeof(struct man_node));
 222         p->line = line;
 223         p->pos = pos;
 224         p->type = type;
 225         p->tok = tok;
 226
 227         if (man->flags & MAN_NEWLINE)
 228                 p->flags |= MAN_LINE;
 229         man->flags &= ~MAN_NEWLINE;
 230         return(p);
 231 }
 232
 233 void
 234 man_elem_alloc(struct man *man, int line, int pos, enum mant tok)
 235 {
 236         struct man_node *p;
 237
 238         p = man_node_alloc(man, line, pos, MAN_ELEM, tok);
 239         man_node_append(man, p);
 240         man->next = MAN_NEXT_CHILD;
 241 }
 242
 243 void
 244 man_head_alloc(struct man *man, int line, int pos, enum mant tok)
 245 {
 246         struct man_node *p;
 247
 248         p = man_node_alloc(man, line, pos, MAN_HEAD, tok);
 249         man_node_append(man, p);
 250         man->next = MAN_NEXT_CHILD;
 251 }
 252
 253 void
 254 man_body_alloc(struct man *man, int line, int pos, enum mant tok)
 255 {
 256         struct man_node *p;
 257
 258         p = man_node_alloc(man, line, pos, MAN_BODY, tok);
 259         man_node_append(man, p);
 260         man->next = MAN_NEXT_CHILD;
 261 }
 262
 263 void
 264 man_block_alloc(struct man *man, int line, int pos, enum mant tok)
 265 {
 266         struct man_node *p;
 267
 268         p = man_node_alloc(man, line, pos, MAN_BLOCK, tok);
 269         man_node_append(man, p);
 270         man->next = MAN_NEXT_CHILD;
 271 }
 272
 273 void
 274 man_word_alloc(struct man *man, int line, int pos, const char *word)
 275 {
 276         struct man_node *n;
 277
 278         n = man_node_alloc(man, line, pos, MAN_TEXT, MAN_MAX);
 279         n->string = roff_strdup(man->roff, word);
 280         man_node_append(man, n);
 281         man->next = MAN_NEXT_SIBLING;
 282 }
 283
 284 void
 285 man_word_append(struct man *man, const char *word)
 286 {
 287         struct man_node *n;
 288         char            *addstr, *newstr;
 289
 290         n = man->last;
 291         addstr = roff_strdup(man->roff, word);
 292         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 293         free(addstr);
 294         free(n->string);
 295         n->string = newstr;
 296         man->next = MAN_NEXT_SIBLING;
 297 }
 298
 299 /*
 300  * Free all of the resources held by a node.  This does NOT unlink a
 301  * node from its context; for that, see man_node_unlink().
 302  */
 303 static void
 304 man_node_free(struct man_node *p)
 305 {
 306
 307         free(p->string);
 308         free(p);
 309 }
 310
 311 void
 312 man_node_delete(struct man *man, struct man_node *p)
 313 {
 314
 315         while (p->child)
 316                 man_node_delete(man, p->child);
 317
 318         man_node_unlink(man, p);
 319         man_node_free(p);
 320 }
 321
 322 void
 323 man_addeqn(struct man *man, const struct eqn *ep)
 324 {
 325         struct man_node *n;
 326
 327         n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 328         n->eqn = ep;
 329         if (ep->ln > man->last->line)
 330                 n->flags |= MAN_LINE;
 331         man_node_append(man, n);
 332         man->next = MAN_NEXT_SIBLING;
 333         man_descope(man, ep->ln, ep->pos);
 334 }
 335
 336 void
 337 man_addspan(struct man *man, const struct tbl_span *sp)
 338 {
 339         struct man_node *n;
 340
 341         n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
 342         n->span = sp;
 343         man_node_append(man, n);
 344         man->next = MAN_NEXT_SIBLING;
 345         man_descope(man, sp->line, 0);
 346 }
 347
 348 static void
 349 man_descope(struct man *man, int line, int offs)
 350 {
 351         /*
 352          * Co-ordinate what happens with having a next-line scope open:
 353          * first close out the element scope (if applicable), then close
 354          * out the block scope (also if applicable).
 355          */
 356
 357         if (man->flags & MAN_ELINE) {
 358                 man->flags &= ~MAN_ELINE;
 359                 man_unscope(man, man->last->parent);
 360         }
 361         if ( ! (man->flags & MAN_BLINE))
 362                 return;
 363         man->flags &= ~MAN_BLINE;
 364         man_unscope(man, man->last->parent);
 365         man_body_alloc(man, line, offs, man->last->tok);
 366 }
 367
 368 static int
 369 man_ptext(struct man *man, int line, char *buf, int offs)
 370 {
 371         int              i;
 372
 373         /* Literal free-form text whitespace is preserved. */
 374
 375         if (man->flags & MAN_LITERAL) {
 376                 man_word_alloc(man, line, offs, buf + offs);
 377                 man_descope(man, line, offs);
 378                 return(1);
 379         }
 380
 381         for (i = offs; buf[i] == ' '; i++)
 382                 /* Skip leading whitespace. */ ;
 383
 384         /*
 385          * Blank lines are ignored right after headings
 386          * but add a single vertical space elsewhere.
 387          */
 388
 389         if (buf[i] == '\0') {
 390                 /* Allocate a blank entry. */
 391                 if (man->last->tok != MAN_SH &&
 392                     man->last->tok != MAN_SS) {
 393                         man_elem_alloc(man, line, offs, MAN_sp);
 394                         man->next = MAN_NEXT_SIBLING;
 395                 }
 396                 return(1);
 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                         mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 410                             line, i - 1, NULL);
 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         man_word_alloc(man, line, offs, buf + offs);
 421
 422         /*
 423          * End-of-sentence check.  If the last character is an unescaped
 424          * EOS character, then flag the node as being the end of a
 425          * sentence.  The front-end will know how to interpret this.
 426          */
 427
 428         assert(i);
 429         if (mandoc_eos(buf, (size_t)i))
 430                 man->last->flags |= MAN_EOS;
 431
 432         man_descope(man, line, offs);
 433         return(1);
 434 }
 435
 436 static int
 437 man_pmacro(struct man *man, int ln, char *buf, int offs)
 438 {
 439         struct man_node *n;
 440         const char      *cp;
 441         enum mant        tok;
 442         int              i, ppos;
 443         int              bline;
 444         char             mac[5];
 445
 446         ppos = offs;
 447
 448         /*
 449          * Copy the first word into a nil-terminated buffer.
 450          * Stop when a space, tab, escape, or eoln is encountered.
 451          */
 452
 453         i = 0;
 454         while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
 455                 mac[i++] = buf[offs++];
 456
 457         mac[i] = '\0';
 458
 459         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 460
 461         if (tok == MAN_MAX) {
 462                 mandoc_msg(MANDOCERR_MACRO, man->parse,
 463                     ln, ppos, buf + ppos - 1);
 464                 return(1);
 465         }
 466
 467         /* Skip a leading escape sequence or tab. */
 468
 469         switch (buf[offs]) {
 470         case '\\':
 471                 cp = buf + offs + 1;
 472                 mandoc_escape(&cp, NULL, NULL);
 473                 offs = cp - buf;
 474                 break;
 475         case '\t':
 476                 offs++;
 477                 break;
 478         default:
 479                 break;
 480         }
 481
 482         /* Jump to the next non-whitespace word. */
 483
 484         while (buf[offs] && buf[offs] == ' ')
 485                 offs++;
 486
 487         /*
 488          * Trailing whitespace.  Note that tabs are allowed to be passed
 489          * into the parser as "text", so we only warn about spaces here.
 490          */
 491
 492         if (buf[offs] == '\0' && buf[offs - 1] == ' ')
 493                 mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 494                     ln, offs - 1, NULL);
 495
 496         /*
 497          * Remove prior ELINE macro, as it's being clobbered by a new
 498          * macro.  Note that NSCOPED macros do not close out ELINE
 499          * macros---they don't print text---so we let those slip by.
 500          */
 501
 502         if ( ! (man_macros[tok].flags & MAN_NSCOPED) &&
 503                         man->flags & MAN_ELINE) {
 504                 n = man->last;
 505                 assert(MAN_TEXT != n->type);
 506
 507                 /* Remove repeated NSCOPED macros causing ELINE. */
 508
 509                 if (man_macros[n->tok].flags & MAN_NSCOPED)
 510                         n = n->parent;
 511
 512                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 513                     n->pos, "%s breaks %s", man_macronames[tok],
 514                     man_macronames[n->tok]);
 515
 516                 man_node_delete(man, n);
 517                 man->flags &= ~MAN_ELINE;
 518         }
 519
 520         /*
 521          * Remove prior BLINE macro that is being clobbered.
 522          */
 523         if ((man->flags & MAN_BLINE) &&
 524             (man_macros[tok].flags & MAN_BSCOPE)) {
 525                 n = man->last;
 526
 527                 /* Might be a text node like 8 in
 528                  * .TP 8
 529                  * .SH foo
 530                  */
 531                 if (n->type == MAN_TEXT)
 532                         n = n->parent;
 533
 534                 /* Remove element that didn't end BLINE, if any. */
 535                 if ( ! (man_macros[n->tok].flags & MAN_BSCOPE))
 536                         n = n->parent;
 537
 538                 assert(n->type == MAN_HEAD);
 539                 n = n->parent;
 540                 assert(n->type == MAN_BLOCK);
 541                 assert(man_macros[n->tok].flags & MAN_SCOPED);
 542
 543                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 544                     n->pos, "%s breaks %s", man_macronames[tok],
 545                     man_macronames[n->tok]);
 546
 547                 man_node_delete(man, n);
 548                 man->flags &= ~MAN_BLINE;
 549         }
 550
 551         /* Remember whether we are in next-line scope for a block head. */
 552
 553         bline = man->flags & MAN_BLINE;
 554
 555         /* Call to handler... */
 556
 557         assert(man_macros[tok].fp);
 558         (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf);
 559
 560         /* In quick mode (for mandocdb), abort after the NAME section. */
 561
 562         if (man->quick && tok == MAN_SH) {
 563                 n = man->last;
 564                 if (n->type == MAN_BODY &&
 565                     strcmp(n->prev->child->string, "NAME"))
 566                         return(2);
 567         }
 568
 569         /*
 570          * If we are in a next-line scope for a block head,
 571          * close it out now and switch to the body,
 572          * unless the next-line scope is allowed to continue.
 573          */
 574
 575         if ( ! bline || man->flags & MAN_ELINE ||
 576             man_macros[tok].flags & MAN_NSCOPED)
 577                 return(1);
 578
 579         assert(man->flags & MAN_BLINE);
 580         man->flags &= ~MAN_BLINE;
 581
 582         man_unscope(man, man->last->parent);
 583         man_body_alloc(man, ln, ppos, man->last->tok);
 584         return(1);
 585 }
 586
 587 /*
 588  * Unlink a node from its context.  If "man" is provided, the last parse
 589  * point will also be adjusted accordingly.
 590  */
 591 static void
 592 man_node_unlink(struct man *man, struct man_node *n)
 593 {
 594
 595         /* Adjust siblings. */
 596
 597         if (n->prev)
 598                 n->prev->next = n->next;
 599         if (n->next)
 600                 n->next->prev = n->prev;
 601
 602         /* Adjust parent. */
 603
 604         if (n->parent) {
 605                 n->parent->nchild--;
 606                 if (n->parent->child == n)
 607                         n->parent->child = n->prev ? n->prev : n->next;
 608         }
 609
 610         /* Adjust parse point, if applicable. */
 611
 612         if (man && man->last == n) {
 613                 /*XXX: this can occur when bailing from validation. */
 614                 /*assert(NULL == n->next);*/
 615                 if (n->prev) {
 616                         man->last = n->prev;
 617                         man->next = MAN_NEXT_SIBLING;
 618                 } else {
 619                         man->last = n->parent;
 620                         man->next = MAN_NEXT_CHILD;
 621                 }
 622         }
 623
 624         if (man && man->first == n)
 625                 man->first = NULL;
 626 }
 627
 628 const struct mparse *
 629 man_mparse(const struct man *man)
 630 {
 631
 632         assert(man && man->parse);
 633         return(man->parse);
 634 }
 635
 636 void
 637 man_deroff(char **dest, const struct man_node *n)
 638 {
 639         char    *cp;
 640         size_t   sz;
 641
 642         if (n->type != MAN_TEXT) {
 643                 for (n = n->child; n; n = n->next)
 644                         man_deroff(dest, n);
 645                 return;
 646         }
 647
 648         /* Skip leading whitespace and escape sequences. */
 649
 650         cp = n->string;
 651         while ('\0' != *cp) {
 652                 if ('\\' == *cp) {
 653                         cp++;
 654                         mandoc_escape((const char **)&cp, NULL, NULL);
 655                 } else if (isspace((unsigned char)*cp))
 656                         cp++;
 657                 else
 658                         break;
 659         }
 660
 661         /* Skip trailing whitespace. */
 662
 663         for (sz = strlen(cp); sz; sz--)
 664                 if (0 == isspace((unsigned char)cp[sz-1]))
 665                         break;
 666
 667         /* Skip empty strings. */
 668
 669         if (0 == sz)
 670                 return;
 671
 672         if (NULL == *dest) {
 673                 *dest = mandoc_strndup(cp, sz);
 674                 return;
 675         }
 676
 677         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 678         free(*dest);
 679         *dest = cp;
 680 }