man.c

   1 /*      $Id: man.c,v 1.154 2015/04/18 16:34:25 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2013, 2014, 2015 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 AUTHORS DISCLAIM ALL WARRANTIES
  12  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  13  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS 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 "mandoc_aux.h"
  31 #include "mandoc.h"
  32 #include "roff.h"
  33 #include "man.h"
  34 #include "libmandoc.h"
  35 #include "libman.h"
  36
  37 const   char *const __man_macronames[MAN_MAX] = {
  38         "br",           "TH",           "SH",           "SS",
  39         "TP",           "LP",           "PP",           "P",
  40         "IP",           "HP",           "SM",           "SB",
  41         "BI",           "IB",           "BR",           "RB",
  42         "R",            "B",            "I",            "IR",
  43         "RI",           "sp",           "nf",
  44         "fi",           "RE",           "RS",           "DT",
  45         "UC",           "PD",           "AT",           "in",
  46         "ft",           "OP",           "EX",           "EE",
  47         "UR",           "UE",           "ll"
  48         };
  49
  50 const   char * const *man_macronames = __man_macronames;
  51
  52 static  void             man_alloc1(struct roff_man *);
  53 static  void             man_breakscope(struct roff_man *, int);
  54 static  void             man_descope(struct roff_man *, int, int);
  55 static  void             man_free1(struct roff_man *);
  56 static  struct roff_node *man_node_alloc(struct roff_man *, int, int,
  57                                 enum roff_type, int);
  58 static  void             man_node_append(struct roff_man *,
  59                                 struct roff_node *);
  60 static  void             man_node_free(struct roff_node *);
  61 static  void             man_node_unlink(struct roff_man *,
  62                                 struct roff_node *);
  63 static  int              man_ptext(struct roff_man *, int, char *, int);
  64 static  int              man_pmacro(struct roff_man *, int, char *, int);
  65
  66
  67 const struct roff_node *
  68 man_node(const struct roff_man *man)
  69 {
  70
  71         return(man->first);
  72 }
  73
  74 const struct roff_meta *
  75 man_meta(const struct roff_man *man)
  76 {
  77
  78         return(&man->meta);
  79 }
  80
  81 void
  82 man_reset(struct roff_man *man)
  83 {
  84
  85         man_free1(man);
  86         man_alloc1(man);
  87 }
  88
  89 void
  90 man_free(struct roff_man *man)
  91 {
  92
  93         man_free1(man);
  94         free(man);
  95 }
  96
  97 struct roff_man *
  98 man_alloc(struct roff *roff, struct mparse *parse,
  99         const char *defos, int quick)
 100 {
 101         struct roff_man *p;
 102
 103         p = mandoc_calloc(1, sizeof(*p));
 104
 105         man_hash_init();
 106         p->parse = parse;
 107         p->defos = defos;
 108         p->quick = quick;
 109         p->roff = roff;
 110
 111         man_alloc1(p);
 112         return(p);
 113 }
 114
 115 void
 116 man_endparse(struct roff_man *man)
 117 {
 118
 119         man_macroend(man);
 120 }
 121
 122 int
 123 man_parseln(struct roff_man *man, int ln, char *buf, int offs)
 124 {
 125
 126         if (man->last->type != ROFFT_EQN || ln > man->last->line)
 127                 man->flags |= MAN_NEWLINE;
 128
 129         return (roff_getcontrol(man->roff, buf, &offs) ?
 130             man_pmacro(man, ln, buf, offs) :
 131             man_ptext(man, ln, buf, offs));
 132 }
 133
 134 static void
 135 man_free1(struct roff_man *man)
 136 {
 137
 138         if (man->first)
 139                 man_node_delete(man, man->first);
 140         free(man->meta.title);
 141         free(man->meta.os);
 142         free(man->meta.date);
 143         free(man->meta.vol);
 144         free(man->meta.msec);
 145 }
 146
 147 static void
 148 man_alloc1(struct roff_man *man)
 149 {
 150
 151         memset(&man->meta, 0, sizeof(man->meta));
 152         man->macroset = MACROSET_MAN;
 153         man->flags = 0;
 154         man->last = mandoc_calloc(1, sizeof(*man->last));
 155         man->first = man->last;
 156         man->last->type = ROFFT_ROOT;
 157         man->last->tok = MAN_MAX;
 158         man->next = ROFF_NEXT_CHILD;
 159 }
 160
 161
 162 static void
 163 man_node_append(struct roff_man *man, struct roff_node *p)
 164 {
 165
 166         assert(man->last);
 167         assert(man->first);
 168         assert(p->type != ROFFT_ROOT);
 169
 170         switch (man->next) {
 171         case ROFF_NEXT_SIBLING:
 172                 man->last->next = p;
 173                 p->prev = man->last;
 174                 p->parent = man->last->parent;
 175                 break;
 176         case ROFF_NEXT_CHILD:
 177                 man->last->child = p;
 178                 p->parent = man->last;
 179                 break;
 180         default:
 181                 abort();
 182                 /* NOTREACHED */
 183         }
 184
 185         assert(p->parent);
 186         p->parent->nchild++;
 187
 188         switch (p->type) {
 189         case ROFFT_BLOCK:
 190                 if (p->tok == MAN_SH || p->tok == MAN_SS)
 191                         man->flags &= ~MAN_LITERAL;
 192                 break;
 193         case ROFFT_HEAD:
 194                 assert(p->parent->type == ROFFT_BLOCK);
 195                 p->parent->head = p;
 196                 break;
 197         case ROFFT_BODY:
 198                 assert(p->parent->type == ROFFT_BLOCK);
 199                 p->parent->body = p;
 200                 break;
 201         default:
 202                 break;
 203         }
 204
 205         man->last = p;
 206
 207         switch (p->type) {
 208         case ROFFT_TBL:
 209                 /* FALLTHROUGH */
 210         case ROFFT_TEXT:
 211                 man_valid_post(man);
 212                 break;
 213         default:
 214                 break;
 215         }
 216 }
 217
 218 static struct roff_node *
 219 man_node_alloc(struct roff_man *man, int line, int pos,
 220                 enum roff_type type, int tok)
 221 {
 222         struct roff_node *p;
 223
 224         p = mandoc_calloc(1, sizeof(*p));
 225         p->line = line;
 226         p->pos = pos;
 227         p->type = type;
 228         p->tok = tok;
 229
 230         if (man->flags & MAN_NEWLINE)
 231                 p->flags |= MAN_LINE;
 232         man->flags &= ~MAN_NEWLINE;
 233         return(p);
 234 }
 235
 236 void
 237 man_elem_alloc(struct roff_man *man, int line, int pos, int tok)
 238 {
 239         struct roff_node *p;
 240
 241         p = man_node_alloc(man, line, pos, ROFFT_ELEM, tok);
 242         man_node_append(man, p);
 243         man->next = ROFF_NEXT_CHILD;
 244 }
 245
 246 void
 247 man_head_alloc(struct roff_man *man, int line, int pos, int tok)
 248 {
 249         struct roff_node *p;
 250
 251         p = man_node_alloc(man, line, pos, ROFFT_HEAD, tok);
 252         man_node_append(man, p);
 253         man->next = ROFF_NEXT_CHILD;
 254 }
 255
 256 void
 257 man_body_alloc(struct roff_man *man, int line, int pos, int tok)
 258 {
 259         struct roff_node *p;
 260
 261         p = man_node_alloc(man, line, pos, ROFFT_BODY, tok);
 262         man_node_append(man, p);
 263         man->next = ROFF_NEXT_CHILD;
 264 }
 265
 266 void
 267 man_block_alloc(struct roff_man *man, int line, int pos, int tok)
 268 {
 269         struct roff_node *p;
 270
 271         p = man_node_alloc(man, line, pos, ROFFT_BLOCK, tok);
 272         man_node_append(man, p);
 273         man->next = ROFF_NEXT_CHILD;
 274 }
 275
 276 void
 277 man_word_alloc(struct roff_man *man, int line, int pos, const char *word)
 278 {
 279         struct roff_node *n;
 280
 281         n = man_node_alloc(man, line, pos, ROFFT_TEXT, MAN_MAX);
 282         n->string = roff_strdup(man->roff, word);
 283         man_node_append(man, n);
 284         man->next = ROFF_NEXT_SIBLING;
 285 }
 286
 287 void
 288 man_word_append(struct roff_man *man, const char *word)
 289 {
 290         struct roff_node *n;
 291         char            *addstr, *newstr;
 292
 293         n = man->last;
 294         addstr = roff_strdup(man->roff, word);
 295         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 296         free(addstr);
 297         free(n->string);
 298         n->string = newstr;
 299         man->next = ROFF_NEXT_SIBLING;
 300 }
 301
 302 /*
 303  * Free all of the resources held by a node.  This does NOT unlink a
 304  * node from its context; for that, see man_node_unlink().
 305  */
 306 static void
 307 man_node_free(struct roff_node *p)
 308 {
 309
 310         free(p->string);
 311         free(p);
 312 }
 313
 314 void
 315 man_node_delete(struct roff_man *man, struct roff_node *p)
 316 {
 317
 318         while (p->child)
 319                 man_node_delete(man, p->child);
 320
 321         man_node_unlink(man, p);
 322         man_node_free(p);
 323 }
 324
 325 void
 326 man_addeqn(struct roff_man *man, const struct eqn *ep)
 327 {
 328         struct roff_node *n;
 329
 330         n = man_node_alloc(man, ep->ln, ep->pos, ROFFT_EQN, MAN_MAX);
 331         n->eqn = ep;
 332         if (ep->ln > man->last->line)
 333                 n->flags |= MAN_LINE;
 334         man_node_append(man, n);
 335         man->next = ROFF_NEXT_SIBLING;
 336         man_descope(man, ep->ln, ep->pos);
 337 }
 338
 339 void
 340 man_addspan(struct roff_man *man, const struct tbl_span *sp)
 341 {
 342         struct roff_node *n;
 343
 344         man_breakscope(man, MAN_MAX);
 345         n = man_node_alloc(man, sp->line, 0, ROFFT_TBL, MAN_MAX);
 346         n->span = sp;
 347         man_node_append(man, n);
 348         man->next = ROFF_NEXT_SIBLING;
 349         man_descope(man, sp->line, 0);
 350 }
 351
 352 static void
 353 man_descope(struct roff_man *man, int line, int offs)
 354 {
 355         /*
 356          * Co-ordinate what happens with having a next-line scope open:
 357          * first close out the element scope (if applicable), then close
 358          * out the block scope (also if applicable).
 359          */
 360
 361         if (man->flags & MAN_ELINE) {
 362                 man->flags &= ~MAN_ELINE;
 363                 man_unscope(man, man->last->parent);
 364         }
 365         if ( ! (man->flags & MAN_BLINE))
 366                 return;
 367         man->flags &= ~MAN_BLINE;
 368         man_unscope(man, man->last->parent);
 369         man_body_alloc(man, line, offs, man->last->tok);
 370 }
 371
 372 static int
 373 man_ptext(struct roff_man *man, int line, char *buf, int offs)
 374 {
 375         int              i;
 376
 377         /* Literal free-form text whitespace is preserved. */
 378
 379         if (man->flags & MAN_LITERAL) {
 380                 man_word_alloc(man, line, offs, buf + offs);
 381                 man_descope(man, line, offs);
 382                 return(1);
 383         }
 384
 385         for (i = offs; buf[i] == ' '; i++)
 386                 /* Skip leading whitespace. */ ;
 387
 388         /*
 389          * Blank lines are ignored right after headings
 390          * but add a single vertical space elsewhere.
 391          */
 392
 393         if (buf[i] == '\0') {
 394                 /* Allocate a blank entry. */
 395                 if (man->last->tok != MAN_SH &&
 396                     man->last->tok != MAN_SS) {
 397                         man_elem_alloc(man, line, offs, MAN_sp);
 398                         man->next = ROFF_NEXT_SIBLING;
 399                 }
 400                 return(1);
 401         }
 402
 403         /*
 404          * Warn if the last un-escaped character is whitespace. Then
 405          * strip away the remaining spaces (tabs stay!).
 406          */
 407
 408         i = (int)strlen(buf);
 409         assert(i);
 410
 411         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 412                 if (i > 1 && '\\' != buf[i - 2])
 413                         mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 414                             line, i - 1, NULL);
 415
 416                 for (--i; i && ' ' == buf[i]; i--)
 417                         /* Spin back to non-space. */ ;
 418
 419                 /* Jump ahead of escaped whitespace. */
 420                 i += '\\' == buf[i] ? 2 : 1;
 421
 422                 buf[i] = '\0';
 423         }
 424         man_word_alloc(man, line, offs, buf + offs);
 425
 426         /*
 427          * End-of-sentence check.  If the last character is an unescaped
 428          * EOS character, then flag the node as being the end of a
 429          * sentence.  The front-end will know how to interpret this.
 430          */
 431
 432         assert(i);
 433         if (mandoc_eos(buf, (size_t)i))
 434                 man->last->flags |= MAN_EOS;
 435
 436         man_descope(man, line, offs);
 437         return(1);
 438 }
 439
 440 static int
 441 man_pmacro(struct roff_man *man, int ln, char *buf, int offs)
 442 {
 443         struct roff_node *n;
 444         const char      *cp;
 445         int              tok;
 446         int              i, ppos;
 447         int              bline;
 448         char             mac[5];
 449
 450         ppos = offs;
 451
 452         /*
 453          * Copy the first word into a nil-terminated buffer.
 454          * Stop when a space, tab, escape, or eoln is encountered.
 455          */
 456
 457         i = 0;
 458         while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
 459                 mac[i++] = buf[offs++];
 460
 461         mac[i] = '\0';
 462
 463         tok = (i > 0 && i < 4) ? man_hash_find(mac) : MAN_MAX;
 464
 465         if (tok == MAN_MAX) {
 466                 mandoc_msg(MANDOCERR_MACRO, man->parse,
 467                     ln, ppos, buf + ppos - 1);
 468                 return(1);
 469         }
 470
 471         /* Skip a leading escape sequence or tab. */
 472
 473         switch (buf[offs]) {
 474         case '\\':
 475                 cp = buf + offs + 1;
 476                 mandoc_escape(&cp, NULL, NULL);
 477                 offs = cp - buf;
 478                 break;
 479         case '\t':
 480                 offs++;
 481                 break;
 482         default:
 483                 break;
 484         }
 485
 486         /* Jump to the next non-whitespace word. */
 487
 488         while (buf[offs] && buf[offs] == ' ')
 489                 offs++;
 490
 491         /*
 492          * Trailing whitespace.  Note that tabs are allowed to be passed
 493          * into the parser as "text", so we only warn about spaces here.
 494          */
 495
 496         if (buf[offs] == '\0' && buf[offs - 1] == ' ')
 497                 mandoc_msg(MANDOCERR_SPACE_EOL, man->parse,
 498                     ln, offs - 1, NULL);
 499
 500         /*
 501          * Some macros break next-line scopes; otherwise, remember
 502          * whether we are in next-line scope for a block head.
 503          */
 504
 505         man_breakscope(man, tok);
 506         bline = man->flags & MAN_BLINE;
 507
 508         /* Call to handler... */
 509
 510         assert(man_macros[tok].fp);
 511         (*man_macros[tok].fp)(man, tok, ln, ppos, &offs, buf);
 512
 513         /* In quick mode (for mandocdb), abort after the NAME section. */
 514
 515         if (man->quick && tok == MAN_SH) {
 516                 n = man->last;
 517                 if (n->type == ROFFT_BODY &&
 518                     strcmp(n->prev->child->string, "NAME"))
 519                         return(2);
 520         }
 521
 522         /*
 523          * If we are in a next-line scope for a block head,
 524          * close it out now and switch to the body,
 525          * unless the next-line scope is allowed to continue.
 526          */
 527
 528         if ( ! bline || man->flags & MAN_ELINE ||
 529             man_macros[tok].flags & MAN_NSCOPED)
 530                 return(1);
 531
 532         assert(man->flags & MAN_BLINE);
 533         man->flags &= ~MAN_BLINE;
 534
 535         man_unscope(man, man->last->parent);
 536         man_body_alloc(man, ln, ppos, man->last->tok);
 537         return(1);
 538 }
 539
 540 void
 541 man_breakscope(struct roff_man *man, int tok)
 542 {
 543         struct roff_node *n;
 544
 545         /*
 546          * An element next line scope is open,
 547          * and the new macro is not allowed inside elements.
 548          * Delete the element that is being broken.
 549          */
 550
 551         if (man->flags & MAN_ELINE && (tok == MAN_MAX ||
 552             ! (man_macros[tok].flags & MAN_NSCOPED))) {
 553                 n = man->last;
 554                 assert(n->type != ROFFT_TEXT);
 555                 if (man_macros[n->tok].flags & MAN_NSCOPED)
 556                         n = n->parent;
 557
 558                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse,
 559                     n->line, n->pos, "%s breaks %s",
 560                     tok == MAN_MAX ? "TS" : man_macronames[tok],
 561                     man_macronames[n->tok]);
 562
 563                 man_node_delete(man, n);
 564                 man->flags &= ~MAN_ELINE;
 565         }
 566
 567         /*
 568          * A block header next line scope is open,
 569          * and the new macro is not allowed inside block headers.
 570          * Delete the block that is being broken.
 571          */
 572
 573         if (man->flags & MAN_BLINE && (tok == MAN_MAX ||
 574             man_macros[tok].flags & MAN_BSCOPE)) {
 575                 n = man->last;
 576                 if (n->type == ROFFT_TEXT)
 577                         n = n->parent;
 578                 if ( ! (man_macros[n->tok].flags & MAN_BSCOPE))
 579                         n = n->parent;
 580
 581                 assert(n->type == ROFFT_HEAD);
 582                 n = n->parent;
 583                 assert(n->type == ROFFT_BLOCK);
 584                 assert(man_macros[n->tok].flags & MAN_SCOPED);
 585
 586                 mandoc_vmsg(MANDOCERR_BLK_LINE, man->parse,
 587                     n->line, n->pos, "%s breaks %s",
 588                     tok == MAN_MAX ? "TS" : man_macronames[tok],
 589                     man_macronames[n->tok]);
 590
 591                 man_node_delete(man, n);
 592                 man->flags &= ~MAN_BLINE;
 593         }
 594 }
 595
 596 /*
 597  * Unlink a node from its context.  If "man" is provided, the last parse
 598  * point will also be adjusted accordingly.
 599  */
 600 static void
 601 man_node_unlink(struct roff_man *man, struct roff_node *n)
 602 {
 603
 604         /* Adjust siblings. */
 605
 606         if (n->prev)
 607                 n->prev->next = n->next;
 608         if (n->next)
 609                 n->next->prev = n->prev;
 610
 611         /* Adjust parent. */
 612
 613         if (n->parent) {
 614                 n->parent->nchild--;
 615                 if (n->parent->child == n)
 616                         n->parent->child = n->prev ? n->prev : n->next;
 617         }
 618
 619         /* Adjust parse point, if applicable. */
 620
 621         if (man && man->last == n) {
 622                 /*XXX: this can occur when bailing from validation. */
 623                 /*assert(NULL == n->next);*/
 624                 if (n->prev) {
 625                         man->last = n->prev;
 626                         man->next = ROFF_NEXT_SIBLING;
 627                 } else {
 628                         man->last = n->parent;
 629                         man->next = ROFF_NEXT_CHILD;
 630                 }
 631         }
 632
 633         if (man && man->first == n)
 634                 man->first = NULL;
 635 }
 636
 637 const struct mparse *
 638 man_mparse(const struct roff_man *man)
 639 {
 640
 641         assert(man && man->parse);
 642         return(man->parse);
 643 }
 644
 645 void
 646 man_deroff(char **dest, const struct roff_node *n)
 647 {
 648         char    *cp;
 649         size_t   sz;
 650
 651         if (n->type != ROFFT_TEXT) {
 652                 for (n = n->child; n; n = n->next)
 653                         man_deroff(dest, n);
 654                 return;
 655         }
 656
 657         /* Skip leading whitespace and escape sequences. */
 658
 659         cp = n->string;
 660         while ('\0' != *cp) {
 661                 if ('\\' == *cp) {
 662                         cp++;
 663                         mandoc_escape((const char **)&cp, NULL, NULL);
 664                 } else if (isspace((unsigned char)*cp))
 665                         cp++;
 666                 else
 667                         break;
 668         }
 669
 670         /* Skip trailing whitespace. */
 671
 672         for (sz = strlen(cp); sz; sz--)
 673                 if (0 == isspace((unsigned char)cp[sz-1]))
 674                         break;
 675
 676         /* Skip empty strings. */
 677
 678         if (0 == sz)
 679                 return;
 680
 681         if (NULL == *dest) {
 682                 *dest = mandoc_strndup(cp, sz);
 683                 return;
 684         }
 685
 686         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 687         free(*dest);
 688         *dest = cp;
 689 }