man.c

   1 /*      $Id: man.c,v 1.51 2010/03/22 14:03:03 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 manual section", /* WMSEC */
  36         "invalid date format", /* WDATE */
  37         "scope of prior line violated", /* WLNSCOPE */
  38         "over-zealous prior line scope violation", /* WLNSCOPE2 */
  39         "trailing whitespace", /* WTSPACE */
  40         "unterminated quoted parameter", /* WTQUOTE */
  41         "document has no body", /* WNODATA */
  42         "document has no title/section", /* WNOTITLE */
  43         "invalid escape sequence", /* WESCAPE */
  44         "invalid number format", /* WNUMFMT */
  45         "expected block head arguments", /* WHEADARGS */
  46         "expected block body arguments", /* WBODYARGS */
  47         "expected empty block head", /* WNHEADARGS */
  48         "ill-formed macro", /* WMACROFORM */
  49         "scope open on exit", /* WEXITSCOPE */
  50         "no scope context", /* WNOSCOPE */
  51         "literal context already open", /* WOLITERAL */
  52         "no literal context open" /* WNLITERAL */
  53 };
  54
  55 const   char *const __man_macronames[MAN_MAX] = {
  56         "br",           "TH",           "SH",           "SS",
  57         "TP",           "LP",           "PP",           "P",
  58         "IP",           "HP",           "SM",           "SB",
  59         "BI",           "IB",           "BR",           "RB",
  60         "R",            "B",            "I",            "IR",
  61         "RI",           "na",           "i",            "sp",
  62         "nf",           "fi",           "r",            "RE",
  63         "RS",           "DT",           "UC",           "PD"
  64         };
  65
  66 const   char * const *man_macronames = __man_macronames;
  67
  68 static  struct man_node *man_node_alloc(int, int,
  69                                 enum man_type, int);
  70 static  int              man_node_append(struct man *,
  71                                 struct man_node *);
  72 static  int              man_ptext(struct man *, int, char *);
  73 static  int              man_pmacro(struct man *, int, char *);
  74 static  void             man_free1(struct man *);
  75 static  void             man_alloc1(struct man *);
  76 static  int              pstring(struct man *, int, int,
  77                                 const char *, size_t);
  78 static  int              macrowarn(struct man *, int, const char *);
  79
  80
  81 const struct man_node *
  82 man_node(const struct man *m)
  83 {
  84
  85         return(MAN_HALT & m->flags ? NULL : m->first);
  86 }
  87
  88
  89 const struct man_meta *
  90 man_meta(const struct man *m)
  91 {
  92
  93         return(MAN_HALT & m->flags ? NULL : &m->meta);
  94 }
  95
  96
  97 void
  98 man_reset(struct man *man)
  99 {
 100
 101         man_free1(man);
 102         man_alloc1(man);
 103 }
 104
 105
 106 void
 107 man_free(struct man *man)
 108 {
 109
 110         man_free1(man);
 111         free(man);
 112 }
 113
 114
 115 struct man *
 116 man_alloc(void *data, int pflags, const struct man_cb *cb)
 117 {
 118         struct man      *p;
 119
 120         p = mandoc_calloc(1, sizeof(struct man));
 121
 122         if (cb)
 123                 memcpy(&p->cb, cb, sizeof(struct man_cb));
 124
 125         man_hash_init();
 126         p->data = data;
 127         p->pflags = pflags;
 128
 129         man_alloc1(p);
 130         return(p);
 131 }
 132
 133
 134 int
 135 man_endparse(struct man *m)
 136 {
 137
 138         if (MAN_HALT & m->flags)
 139                 return(0);
 140         else if (man_macroend(m))
 141                 return(1);
 142         m->flags |= MAN_HALT;
 143         return(0);
 144 }
 145
 146
 147 int
 148 man_parseln(struct man *m, int ln, char *buf)
 149 {
 150
 151         return('.' == *buf ?
 152                         man_pmacro(m, ln, buf) :
 153                         man_ptext(m, ln, buf));
 154 }
 155
 156
 157 static void
 158 man_free1(struct man *man)
 159 {
 160
 161         if (man->first)
 162                 man_node_freelist(man->first);
 163         if (man->meta.title)
 164                 free(man->meta.title);
 165         if (man->meta.source)
 166                 free(man->meta.source);
 167         if (man->meta.vol)
 168                 free(man->meta.vol);
 169 }
 170
 171
 172 static void
 173 man_alloc1(struct man *m)
 174 {
 175
 176         memset(&m->meta, 0, sizeof(struct man_meta));
 177         m->flags = 0;
 178         m->last = mandoc_calloc(1, sizeof(struct man_node));
 179         m->first = m->last;
 180         m->last->type = MAN_ROOT;
 181         m->next = MAN_NEXT_CHILD;
 182 }
 183
 184
 185 static int
 186 man_node_append(struct man *man, struct man_node *p)
 187 {
 188
 189         assert(man->last);
 190         assert(man->first);
 191         assert(MAN_ROOT != p->type);
 192
 193         switch (man->next) {
 194         case (MAN_NEXT_SIBLING):
 195                 man->last->next = p;
 196                 p->prev = man->last;
 197                 p->parent = man->last->parent;
 198                 break;
 199         case (MAN_NEXT_CHILD):
 200                 man->last->child = p;
 201                 p->parent = man->last;
 202                 break;
 203         default:
 204                 abort();
 205                 /* NOTREACHED */
 206         }
 207
 208         p->parent->nchild++;
 209
 210         if ( ! man_valid_pre(man, p))
 211                 return(0);
 212
 213         switch (p->type) {
 214         case (MAN_HEAD):
 215                 assert(MAN_BLOCK == p->parent->type);
 216                 p->parent->head = p;
 217                 break;
 218         case (MAN_BODY):
 219                 assert(MAN_BLOCK == p->parent->type);
 220                 p->parent->body = p;
 221                 break;
 222         default:
 223                 break;
 224         }
 225
 226         man->last = p;
 227
 228         switch (p->type) {
 229         case (MAN_TEXT):
 230                 if ( ! man_valid_post(man))
 231                         return(0);
 232                 if ( ! man_action_post(man))
 233                         return(0);
 234                 break;
 235         default:
 236                 break;
 237         }
 238
 239         return(1);
 240 }
 241
 242
 243 static struct man_node *
 244 man_node_alloc(int line, int pos, enum man_type type, int tok)
 245 {
 246         struct man_node *p;
 247
 248         p = mandoc_calloc(1, sizeof(struct man_node));
 249         p->line = line;
 250         p->pos = pos;
 251         p->type = type;
 252         p->tok = tok;
 253         return(p);
 254 }
 255
 256
 257 int
 258 man_elem_alloc(struct man *m, int line, int pos, int tok)
 259 {
 260         struct man_node *p;
 261
 262         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 263         if ( ! man_node_append(m, p))
 264                 return(0);
 265         m->next = MAN_NEXT_CHILD;
 266         return(1);
 267 }
 268
 269
 270 int
 271 man_head_alloc(struct man *m, int line, int pos, int tok)
 272 {
 273         struct man_node *p;
 274
 275         p = man_node_alloc(line, pos, MAN_HEAD, tok);
 276         if ( ! man_node_append(m, p))
 277                 return(0);
 278         m->next = MAN_NEXT_CHILD;
 279         return(1);
 280 }
 281
 282
 283 int
 284 man_body_alloc(struct man *m, int line, int pos, int tok)
 285 {
 286         struct man_node *p;
 287
 288         p = man_node_alloc(line, pos, MAN_BODY, tok);
 289         if ( ! man_node_append(m, p))
 290                 return(0);
 291         m->next = MAN_NEXT_CHILD;
 292         return(1);
 293 }
 294
 295
 296 int
 297 man_block_alloc(struct man *m, int line, int pos, int tok)
 298 {
 299         struct man_node *p;
 300
 301         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 302         if ( ! man_node_append(m, p))
 303                 return(0);
 304         m->next = MAN_NEXT_CHILD;
 305         return(1);
 306 }
 307
 308
 309 static int
 310 pstring(struct man *m, int line, int pos,
 311                 const char *p, size_t len)
 312 {
 313         struct man_node *n;
 314         size_t           sv;
 315
 316         n = man_node_alloc(line, pos, MAN_TEXT, -1);
 317         n->string = mandoc_malloc(len + 1);
 318         sv = strlcpy(n->string, p, len + 1);
 319
 320         /* Prohibit truncation. */
 321         assert(sv < len + 1);
 322
 323         if ( ! man_node_append(m, n))
 324                 return(0);
 325         m->next = MAN_NEXT_SIBLING;
 326         return(1);
 327 }
 328
 329
 330 int
 331 man_word_alloc(struct man *m, int line, int pos, const char *word)
 332 {
 333
 334         return(pstring(m, line, pos, word, strlen(word)));
 335 }
 336
 337
 338 void
 339 man_node_free(struct man_node *p)
 340 {
 341
 342         if (p->string)
 343                 free(p->string);
 344         if (p->parent)
 345                 p->parent->nchild--;
 346         free(p);
 347 }
 348
 349
 350 void
 351 man_node_freelist(struct man_node *p)
 352 {
 353         struct man_node *n;
 354
 355         if (p->child)
 356                 man_node_freelist(p->child);
 357         assert(0 == p->nchild);
 358         n = p->next;
 359         man_node_free(p);
 360         if (n)
 361                 man_node_freelist(n);
 362 }
 363
 364
 365 static int
 366 man_ptext(struct man *m, int line, char *buf)
 367 {
 368         int              i, j;
 369         char             sv;
 370
 371         /* Literal free-form text whitespace is preserved. */
 372
 373         if (MAN_LITERAL & m->flags) {
 374                 if ( ! man_word_alloc(m, line, 0, buf))
 375                         return(0);
 376                 goto descope;
 377         }
 378
 379         /* First de-chunk and allocate words. */
 380
 381         for (i = 0; ' ' == buf[i]; i++)
 382                 /* Skip leading whitespace. */ ;
 383
 384         if ('\0' == buf[i]) {
 385                 /* Trailing whitespace? */
 386                 if (i && ' ' == buf[i - 1])
 387                         if ( ! man_pwarn(m, line, i - 1, WTSPACE))
 388                                 return(0);
 389                 if ( ! pstring(m, line, 0, &buf[i], 0))
 390                         return(0);
 391                 goto descope;
 392         }
 393
 394         for (j = i; buf[i]; i++) {
 395                 if (' ' != buf[i])
 396                         continue;
 397
 398                 /* Escaped whitespace. */
 399                 if (i && ' ' == buf[i] && '\\' == buf[i - 1])
 400                         continue;
 401
 402                 sv = buf[i];
 403                 buf[i++] = '\0';
 404
 405                 if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 406                         return(0);
 407
 408                 /* Trailing whitespace?  Check at overwritten byte. */
 409
 410                 if (' ' == sv && '\0' == buf[i])
 411                         if ( ! man_pwarn(m, line, i - 1, WTSPACE))
 412                                 return(0);
 413
 414                 for ( ; ' ' == buf[i]; i++)
 415                         /* Skip trailing whitespace. */ ;
 416
 417                 j = i;
 418
 419                 /* Trailing whitespace? */
 420
 421                 if (' ' == buf[i - 1] && '\0' == buf[i])
 422                         if ( ! man_pwarn(m, line, i - 1, WTSPACE))
 423                                 return(0);
 424
 425                 if ('\0' == buf[i])
 426                         break;
 427         }
 428
 429         if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 430                 return(0);
 431
 432 descope:
 433
 434         /*
 435          * Co-ordinate what happens with having a next-line scope open:
 436          * first close out the element scope (if applicable), then close
 437          * out the block scope (also if applicable).
 438          */
 439
 440         if (MAN_ELINE & m->flags) {
 441                 m->flags &= ~MAN_ELINE;
 442                 if ( ! man_unscope(m, m->last->parent))
 443                         return(0);
 444         }
 445
 446         if ( ! (MAN_BLINE & m->flags))
 447                 return(1);
 448         m->flags &= ~MAN_BLINE;
 449
 450         if ( ! man_unscope(m, m->last->parent))
 451                 return(0);
 452         return(man_body_alloc(m, line, 0, m->last->tok));
 453 }
 454
 455
 456 static int
 457 macrowarn(struct man *m, int ln, const char *buf)
 458 {
 459         if ( ! (MAN_IGN_MACRO & m->pflags))
 460                 return(man_verr(m, ln, 0,
 461                                 "unknown macro: %s%s",
 462                                 buf, strlen(buf) > 3 ? "..." : ""));
 463         return(man_vwarn(m, ln, 0, "unknown macro: %s%s",
 464                                 buf, strlen(buf) > 3 ? "..." : ""));
 465 }
 466
 467
 468 int
 469 man_pmacro(struct man *m, int ln, char *buf)
 470 {
 471         int              i, j, c, ppos, fl;
 472         char             mac[5];
 473         struct man_node *n;
 474
 475         /* Comments and empties are quickly ignored. */
 476
 477         fl = m->flags;
 478
 479         if ('\0' == buf[1])
 480                 return(1);
 481
 482         i = 1;
 483
 484         if (' ' == buf[i]) {
 485                 i++;
 486                 while (buf[i] && ' ' == buf[i])
 487                         i++;
 488                 if ('\0' == buf[i])
 489                         goto out;
 490         }
 491
 492         ppos = i;
 493
 494         /* Copy the first word into a nil-terminated buffer. */
 495
 496         for (j = 0; j < 4; j++, i++) {
 497                 if ('\0' == (mac[j] = buf[i]))
 498                         break;
 499                 else if (' ' == buf[i])
 500                         break;
 501
 502                 /* Check for invalid characters. */
 503
 504                 if (isgraph((u_char)buf[i]))
 505                         continue;
 506                 return(man_perr(m, ln, i, WNPRINT));
 507         }
 508
 509         mac[j] = '\0';
 510
 511         if (j == 4 || j < 1) {
 512                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 513                         (void)man_perr(m, ln, ppos, WMACROFORM);
 514                         goto err;
 515                 }
 516                 if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
 517                         goto err;
 518                 return(1);
 519         }
 520
 521         if (MAN_MAX == (c = man_hash_find(mac))) {
 522                 if ( ! macrowarn(m, ln, mac))
 523                         goto err;
 524                 return(1);
 525         }
 526
 527         /* The macro is sane.  Jump to the next word. */
 528
 529         while (buf[i] && ' ' == buf[i])
 530                 i++;
 531
 532         /* Trailing whitespace? */
 533
 534         if ('\0' == buf[i] && ' ' == buf[i - 1])
 535                 if ( ! man_pwarn(m, ln, i - 1, WTSPACE))
 536                         goto err;
 537
 538         /*
 539          * Remove prior ELINE macro, as it's being clobbering by a new
 540          * macro.  Note that NSCOPED macros do not close out ELINE
 541          * macros---they don't print text---so we let those slip by.
 542          */
 543
 544         if ( ! (MAN_NSCOPED & man_macros[c].flags) &&
 545                         m->flags & MAN_ELINE) {
 546                 assert(MAN_TEXT != m->last->type);
 547
 548                 /*
 549                  * This occurs in the following construction:
 550                  *   .B
 551                  *   .br
 552                  *   .B
 553                  *   .br
 554                  *   I hate man macros.
 555                  * Flat-out disallow this madness.
 556                  */
 557                 if (MAN_NSCOPED & man_macros[m->last->tok].flags)
 558                         return(man_perr(m, ln, ppos, WLNSCOPE));
 559
 560                 n = m->last;
 561
 562                 assert(n);
 563                 assert(NULL == n->child);
 564                 assert(0 == n->nchild);
 565
 566                 if ( ! man_nwarn(m, n, WLNSCOPE))
 567                         return(0);
 568
 569                 man_node_unlink(m, n);
 570                 man_node_free(n);
 571                 m->flags &= ~MAN_ELINE;
 572         }
 573
 574         /* Begin recursive parse sequence. */
 575
 576         assert(man_macros[c].fp);
 577
 578         if ( ! (*man_macros[c].fp)(m, c, ln, ppos, &i, buf))
 579                 goto err;
 580
 581 out:
 582         /*
 583          * We weren't in a block-line scope when entering the
 584          * above-parsed macro, so return.
 585          */
 586
 587         if ( ! (MAN_BLINE & fl)) {
 588                 m->flags &= ~MAN_ILINE;
 589                 return(1);
 590         }
 591
 592         /*
 593          * If we're in a block scope, then allow this macro to slip by
 594          * without closing scope around it.
 595          */
 596
 597         if (MAN_ILINE & m->flags) {
 598                 m->flags &= ~MAN_ILINE;
 599                 return(1);
 600         }
 601
 602         /*
 603          * If we've opened a new next-line element scope, then return
 604          * now, as the next line will close out the block scope.
 605          */
 606
 607         if (MAN_ELINE & m->flags)
 608                 return(1);
 609
 610         /* Close out the block scope opened in the prior line.  */
 611
 612         assert(MAN_BLINE & m->flags);
 613         m->flags &= ~MAN_BLINE;
 614
 615         if ( ! man_unscope(m, m->last->parent))
 616                 return(0);
 617         return(man_body_alloc(m, ln, 0, m->last->tok));
 618
 619 err:    /* Error out. */
 620
 621         m->flags |= MAN_HALT;
 622         return(0);
 623 }
 624
 625
 626 int
 627 man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
 628 {
 629         char             buf[256];
 630         va_list          ap;
 631
 632         if (NULL == man->cb.man_err)
 633                 return(0);
 634
 635         va_start(ap, fmt);
 636         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 637         va_end(ap);
 638         return((*man->cb.man_err)(man->data, ln, pos, buf));
 639 }
 640
 641
 642 int
 643 man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 644 {
 645         char             buf[256];
 646         va_list          ap;
 647
 648         if (NULL == man->cb.man_warn)
 649                 return(0);
 650
 651         va_start(ap, fmt);
 652         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 653         va_end(ap);
 654         return((*man->cb.man_warn)(man->data, ln, pos, buf));
 655 }
 656
 657
 658 int
 659 man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 660 {
 661         const char       *p;
 662
 663         p = __man_merrnames[(int)type];
 664         assert(p);
 665
 666         if (iserr)
 667                 return(man_verr(m, line, pos, p));
 668
 669         return(man_vwarn(m, line, pos, p));
 670 }
 671
 672
 673 void
 674 man_node_unlink(struct man *m, struct man_node *n)
 675 {
 676
 677         if (n->prev) {
 678                 n->prev->next = n->next;
 679                 if (m->last == n) {
 680                         assert(NULL == n->next);
 681                         m->last = n->prev;
 682                         m->next = MAN_NEXT_SIBLING;
 683                 }
 684         } else {
 685                 n->parent->child = n->next;
 686                 if (m->last == n) {
 687                         assert(NULL == n->next);
 688                         m->last = n->parent;
 689                         m->next = MAN_NEXT_CHILD;
 690                 }
 691         }
 692
 693         if (n->next)
 694                 n->next->prev = n->prev;
 695 }