man.c

   1 /*      $Id: man.c,v 1.147 2015/01/15 04:26:40 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 void
 113 man_endparse(struct man *man)
 114 {
 115
 116         man_macroend(man);
 117 }
 118
 119 int
 120 man_parseln(struct man *man, int ln, char *buf, int offs)
 121 {
 122
 123         if (man->last->type != MAN_EQN || ln > man->last->line)
 124                 man->flags |= MAN_NEWLINE;
 125
 126         return (roff_getcontrol(man->roff, buf, &offs) ?
 127             man_pmacro(man, ln, buf, offs) :
 128             man_ptext(man, ln, buf, offs));
 129 }
 130
 131 static void
 132 man_free1(struct man *man)
 133 {
 134
 135         if (man->first)
 136                 man_node_delete(man, man->first);
 137         free(man->meta.title);
 138         free(man->meta.source);
 139         free(man->meta.date);
 140         free(man->meta.vol);
 141         free(man->meta.msec);
 142 }
 143
 144 static void
 145 man_alloc1(struct man *man)
 146 {
 147
 148         memset(&man->meta, 0, sizeof(struct man_meta));
 149         man->flags = 0;
 150         man->last = mandoc_calloc(1, sizeof(struct man_node));
 151         man->first = man->last;
 152         man->last->type = MAN_ROOT;
 153         man->last->tok = MAN_MAX;
 154         man->next = MAN_NEXT_CHILD;
 155 }
 156
 157
 158 static void
 159 man_node_append(struct man *man, struct man_node *p)
 160 {
 161
 162         assert(man->last);
 163         assert(man->first);
 164         assert(p->type != MAN_ROOT);
 165
 166         switch (man->next) {
 167         case MAN_NEXT_SIBLING:
 168                 man->last->next = p;
 169                 p->prev = man->last;
 170                 p->parent = man->last->parent;
 171                 break;
 172         case MAN_NEXT_CHILD:
 173                 man->last->child = p;
 174                 p->parent = man->last;
 175                 break;
 176         default:
 177                 abort();
 178                 /* NOTREACHED */
 179         }
 180
 181         assert(p->parent);
 182         p->parent->nchild++;
 183
 184         switch (p->type) {
 185         case MAN_BLOCK:
 186                 if (p->tok == MAN_SH || p->tok == MAN_SS)
 187                         man->flags &= ~MAN_LITERAL;
 188                 break;
 189         case MAN_HEAD:
 190                 assert(p->parent->type == MAN_BLOCK);
 191                 p->parent->head = p;
 192                 break;
 193         case MAN_BODY:
 194                 assert(p->parent->type == MAN_BLOCK);
 195                 p->parent->body = p;
 196                 break;
 197         default:
 198                 break;
 199         }
 200
 201         man->last = p;
 202
 203         switch (p->type) {
 204         case MAN_TBL:
 205                 /* FALLTHROUGH */
 206         case MAN_TEXT:
 207                 man_valid_post(man);
 208                 break;
 209         default:
 210                 break;
 211         }
 212 }
 213
 214 static struct man_node *
 215 man_node_alloc(struct man *man, int line, int pos,
 216                 enum man_type type, enum mant tok)
 217 {
 218         struct man_node *p;
 219
 220         p = mandoc_calloc(1, sizeof(struct man_node));
 221         p->line = line;
 222         p->pos = pos;
 223         p->type = type;
 224         p->tok = tok;
 225
 226         if (man->flags & MAN_NEWLINE)
 227                 p->flags |= MAN_LINE;
 228         man->flags &= ~MAN_NEWLINE;
 229         return(p);
 230 }
 231
 232 void
 233 man_elem_alloc(struct man *man, int line, int pos, enum mant tok)
 234 {
 235         struct man_node *p;
 236
 237         p = man_node_alloc(man, line, pos, MAN_ELEM, tok);
 238         man_node_append(man, p);
 239         man->next = MAN_NEXT_CHILD;
 240 }
 241
 242 void
 243 man_head_alloc(struct man *man, int line, int pos, enum mant tok)
 244 {
 245         struct man_node *p;
 246
 247         p = man_node_alloc(man, line, pos, MAN_HEAD, tok);
 248         man_node_append(man, p);
 249         man->next = MAN_NEXT_CHILD;
 250 }
 251
 252 void
 253 man_body_alloc(struct man *man, int line, int pos, enum mant tok)
 254 {
 255         struct man_node *p;
 256
 257         p = man_node_alloc(man, line, pos, MAN_BODY, tok);
 258         man_node_append(man, p);
 259         man->next = MAN_NEXT_CHILD;
 260 }
 261
 262 void
 263 man_block_alloc(struct man *man, int line, int pos, enum mant tok)
 264 {
 265         struct man_node *p;
 266
 267         p = man_node_alloc(man, line, pos, MAN_BLOCK, tok);
 268         man_node_append(man, p);
 269         man->next = MAN_NEXT_CHILD;
 270 }
 271
 272 void
 273 man_word_alloc(struct man *man, int line, int pos, const char *word)
 274 {
 275         struct man_node *n;
 276
 277         n = man_node_alloc(man, line, pos, MAN_TEXT, MAN_MAX);
 278         n->string = roff_strdup(man->roff, word);
 279         man_node_append(man, n);
 280         man->next = MAN_NEXT_SIBLING;
 281 }
 282
 283 void
 284 man_word_append(struct man *man, const char *word)
 285 {
 286         struct man_node *n;
 287         char            *addstr, *newstr;
 288
 289         n = man->last;
 290         addstr = roff_strdup(man->roff, word);
 291         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 292         free(addstr);
 293         free(n->string);
 294         n->string = newstr;
 295         man->next = MAN_NEXT_SIBLING;
 296 }
 297
 298 /*
 299  * Free all of the resources held by a node.  This does NOT unlink a
 300  * node from its context; for that, see man_node_unlink().
 301  */
 302 static void
 303 man_node_free(struct man_node *p)
 304 {
 305
 306         free(p->string);
 307         free(p);
 308 }
 309
 310 void
 311 man_node_delete(struct man *man, struct man_node *p)
 312 {
 313
 314         while (p->child)
 315                 man_node_delete(man, p->child);
 316
 317         man_node_unlink(man, p);
 318         man_node_free(p);
 319 }
 320
 321 void
 322 man_addeqn(struct man *man, const struct eqn *ep)
 323 {
 324         struct man_node *n;
 325
 326         n = man_node_alloc(man, ep->ln, ep->pos, MAN_EQN, MAN_MAX);
 327         n->eqn = ep;
 328         if (ep->ln > man->last->line)
 329                 n->flags |= MAN_LINE;
 330         man_node_append(man, n);
 331         man->next = MAN_NEXT_SIBLING;
 332         man_descope(man, ep->ln, ep->pos);
 333 }
 334
 335 void
 336 man_addspan(struct man *man, const struct tbl_span *sp)
 337 {
 338         struct man_node *n;
 339
 340         n = man_node_alloc(man, sp->line, 0, MAN_TBL, MAN_MAX);
 341         n->span = sp;
 342         man_node_append(man, n);
 343         man->next = MAN_NEXT_SIBLING;
 344         man_descope(man, sp->line, 0);
 345 }
 346
 347 static void
 348 man_descope(struct man *man, int line, int offs)
 349 {
 350         /*
 351          * Co-ordinate what happens with having a next-line scope open:
 352          * first close out the element scope (if applicable), then close
 353          * out the block scope (also if applicable).
 354          */
 355
 356         if (man->flags & MAN_ELINE) {
 357                 man->flags &= ~MAN_ELINE;
 358                 man_unscope(man, man->last->parent);
 359         }
 360         if ( ! (man->flags & MAN_BLINE))
 361                 return;
 362         man->flags &= ~MAN_BLINE;
 363         man_unscope(man, man->last->parent);
 364         man_body_alloc(man, line, offs, man->last->tok);
 365 }
 366
 367 static int
 368 man_ptext(struct man *man, int line, char *buf, int offs)
 369 {
 370         int              i;
 371
 372         /* Literal free-form text whitespace is preserved. */
 373
 374         if (man->flags & MAN_LITERAL) {
 375                 man_word_alloc(man, line, offs, buf + offs);
 376                 man_descope(man, line, offs);
 377                 return(1);
 378         }
 379
 380         for (i = offs; buf[i] == ' '; i++)
 381                 /* Skip leading whitespace. */ ;
 382
 383         /*
 384          * Blank lines are ignored right after headings
 385          * but add a single vertical space elsewhere.
 386          */
 387
 388         if (buf[i] == '\0') {
 389                 /* Allocate a blank entry. */
 390                 if (man->last->tok != MAN_SH &&
 391                     man->last->tok != MAN_SS) {
 392                         man_elem_alloc(man, line, offs, MAN_sp);
 393                         man->next = MAN_NEXT_SIBLING;
 394                 }
 395                 return(1);
 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                         mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 409                             line, i - 1, NULL);
 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         man_word_alloc(man, line, offs, buf + offs);
 420
 421         /*
 422          * End-of-sentence check.  If the last character is an unescaped
 423          * EOS character, then flag the node as being the end of a
 424          * sentence.  The front-end will know how to interpret this.
 425          */
 426
 427         assert(i);
 428         if (mandoc_eos(buf, (size_t)i))
 429                 man->last->flags |= MAN_EOS;
 430
 431         man_descope(man, line, offs);
 432         return(1);
 433 }
 434
 435 static int
 436 man_pmacro(struct man *man, int ln, char *buf, int offs)
 437 {
 438         struct man_node *n;
 439         const char      *cp;
 440         enum mant        tok;
 441         int              i, ppos;
 442         int              bline;
 443         char             mac[5];
 444
 445         ppos = offs;
 446
 447         /*
 448          * Copy the first word into a nil-terminated buffer.
 449          * Stop when a space, tab, escape, or eoln is encountered.
 450          */
 451
 452         i = 0;
 453         while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
 454                 mac[i++] = buf[offs++];
 455
 456         mac[i] = '\0';
 457
 458         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 459
 460         if (tok == MAN_MAX) {
 461                 mandoc_msg(MANDOCERR_MACRO, man->parse,
 462                     ln, ppos, buf + ppos - 1);
 463                 return(1);
 464         }
 465
 466         /* Skip a leading escape sequence or tab. */
 467
 468         switch (buf[offs]) {
 469         case '\\':
 470                 cp = buf + offs + 1;
 471                 mandoc_escape(&cp, NULL, NULL);
 472                 offs = cp - buf;
 473                 break;
 474         case '\t':
 475                 offs++;
 476                 break;
 477         default:
 478                 break;
 479         }
 480
 481         /* Jump to the next non-whitespace word. */
 482
 483         while (buf[offs] && buf[offs] == ' ')
 484                 offs++;
 485
 486         /*
 487          * Trailing whitespace.  Note that tabs are allowed to be passed
 488          * into the parser as "text", so we only warn about spaces here.
 489          */
 490
 491         if (buf[offs] == '\0' && buf[offs - 1] == ' ')
 492                 mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 493                     ln, offs - 1, NULL);
 494
 495         /*
 496          * Remove prior ELINE macro, as it's being clobbered by a new
 497          * macro.  Note that NSCOPED macros do not close out ELINE
 498          * macros---they don't print text---so we let those slip by.
 499          */
 500
 501         if ( ! (man_macros[tok].flags & MAN_NSCOPED) &&
 502                         man->flags & MAN_ELINE) {
 503                 n = man->last;
 504                 assert(MAN_TEXT != n->type);
 505
 506                 /* Remove repeated NSCOPED macros causing ELINE. */
 507
 508                 if (man_macros[n->tok].flags & MAN_NSCOPED)
 509                         n = n->parent;
 510
 511                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 512                     n->pos, "%s breaks %s", man_macronames[tok],
 513                     man_macronames[n->tok]);
 514
 515                 man_node_delete(man, n);
 516                 man->flags &= ~MAN_ELINE;
 517         }
 518
 519         /*
 520          * Remove prior BLINE macro that is being clobbered.
 521          */
 522         if ((man->flags & MAN_BLINE) &&
 523             (man_macros[tok].flags & MAN_BSCOPE)) {
 524                 n = man->last;
 525
 526                 /* Might be a text node like 8 in
 527                  * .TP 8
 528                  * .SH foo
 529                  */
 530                 if (n->type == MAN_TEXT)
 531                         n = n->parent;
 532
 533                 /* Remove element that didn't end BLINE, if any. */
 534                 if ( ! (man_macros[n->tok].flags & MAN_BSCOPE))
 535                         n = n->parent;
 536
 537                 assert(n->type == MAN_HEAD);
 538                 n = n->parent;
 539                 assert(n->type == MAN_BLOCK);
 540                 assert(man_macros[n->tok].flags & MAN_SCOPED);
 541
 542                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse, n->line,
 543                     n->pos, "%s breaks %s", man_macronames[tok],
 544                     man_macronames[n->tok]);
 545
 546                 man_node_delete(man, n);
 547                 man->flags &= ~MAN_BLINE;
 548         }
 549
 550         /* Remember whether we are in next-line scope for a block head. */
 551
 552         bline = man->flags & MAN_BLINE;
 553
 554         /* Call to handler... */
 555
 556         assert(man_macros[tok].fp);
 557         (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf);
 558
 559         /* In quick mode (for mandocdb), abort after the NAME section. */
 560
 561         if (man->quick && tok == MAN_SH) {
 562                 n = man->last;
 563                 if (n->type == MAN_BODY &&
 564                     strcmp(n->prev->child->string, "NAME"))
 565                         return(2);
 566         }
 567
 568         /*
 569          * If we are in a next-line scope for a block head,
 570          * close it out now and switch to the body,
 571          * unless the next-line scope is allowed to continue.
 572          */
 573
 574         if ( ! bline || man->flags & MAN_ELINE ||
 575             man_macros[tok].flags & MAN_NSCOPED)
 576                 return(1);
 577
 578         assert(man->flags & MAN_BLINE);
 579         man->flags &= ~MAN_BLINE;
 580
 581         man_unscope(man, man->last->parent);
 582         man_body_alloc(man, ln, ppos, man->last->tok);
 583         return(1);
 584 }
 585
 586 /*
 587  * Unlink a node from its context.  If "man" is provided, the last parse
 588  * point will also be adjusted accordingly.
 589  */
 590 static void
 591 man_node_unlink(struct man *man, struct man_node *n)
 592 {
 593
 594         /* Adjust siblings. */
 595
 596         if (n->prev)
 597                 n->prev->next = n->next;
 598         if (n->next)
 599                 n->next->prev = n->prev;
 600
 601         /* Adjust parent. */
 602
 603         if (n->parent) {
 604                 n->parent->nchild--;
 605                 if (n->parent->child == n)
 606                         n->parent->child = n->prev ? n->prev : n->next;
 607         }
 608
 609         /* Adjust parse point, if applicable. */
 610
 611         if (man && man->last == n) {
 612                 /*XXX: this can occur when bailing from validation. */
 613                 /*assert(NULL == n->next);*/
 614                 if (n->prev) {
 615                         man->last = n->prev;
 616                         man->next = MAN_NEXT_SIBLING;
 617                 } else {
 618                         man->last = n->parent;
 619                         man->next = MAN_NEXT_CHILD;
 620                 }
 621         }
 622
 623         if (man && man->first == n)
 624                 man->first = NULL;
 625 }
 626
 627 const struct mparse *
 628 man_mparse(const struct man *man)
 629 {
 630
 631         assert(man && man->parse);
 632         return(man->parse);
 633 }
 634
 635 void
 636 man_deroff(char **dest, const struct man_node *n)
 637 {
 638         char    *cp;
 639         size_t   sz;
 640
 641         if (n->type != MAN_TEXT) {
 642                 for (n = n->child; n; n = n->next)
 643                         man_deroff(dest, n);
 644                 return;
 645         }
 646
 647         /* Skip leading whitespace and escape sequences. */
 648
 649         cp = n->string;
 650         while ('\0' != *cp) {
 651                 if ('\\' == *cp) {
 652                         cp++;
 653                         mandoc_escape((const char **)&cp, NULL, NULL);
 654                 } else if (isspace((unsigned char)*cp))
 655                         cp++;
 656                 else
 657                         break;
 658         }
 659
 660         /* Skip trailing whitespace. */
 661
 662         for (sz = strlen(cp); sz; sz--)
 663                 if (0 == isspace((unsigned char)cp[sz-1]))
 664                         break;
 665
 666         /* Skip empty strings. */
 667
 668         if (0 == sz)
 669                 return;
 670
 671         if (NULL == *dest) {
 672                 *dest = mandoc_strndup(cp, sz);
 673                 return;
 674         }
 675
 676         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 677         free(*dest);
 678         *dest = cp;
 679 }