man.c

   1 /*      $Id: man.c,v 1.71 2010/05/16 00:04:46 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 date format", /* WDATE */
  36         "scope of prior line violated", /* WLNSCOPE */
  37         "over-zealous prior line scope violation", /* WLNSCOPE2 */
  38         "trailing whitespace", /* WTSPACE */
  39         "unterminated quoted parameter", /* WTQUOTE */
  40         "document has no body", /* WNODATA */
  41         "document has no title/section", /* WNOTITLE */
  42         "invalid escape sequence", /* WESCAPE */
  43         "invalid number format", /* WNUMFMT */
  44         "expected block head arguments", /* WHEADARGS */
  45         "expected block body arguments", /* WBODYARGS */
  46         "expected empty block head", /* WNHEADARGS */
  47         "ill-formed macro", /* WMACROFORM */
  48         "scope open on exit", /* WEXITSCOPE */
  49         "no scope context", /* WNOSCOPE */
  50         "literal context already open", /* WOLITERAL */
  51         "no literal context open", /* WNLITERAL */
  52         "document title should be uppercase", /* WTITLECASE */
  53         "deprecated comment style", /* WBADCOMMENT */
  54 };
  55
  56 const   char *const __man_macronames[MAN_MAX] = {
  57         "br",           "TH",           "SH",           "SS",
  58         "TP",           "LP",           "PP",           "P",
  59         "IP",           "HP",           "SM",           "SB",
  60         "BI",           "IB",           "BR",           "RB",
  61         "R",            "B",            "I",            "IR",
  62         "RI",           "na",           "i",            "sp",
  63         "nf",           "fi",           "r",            "RE",
  64         "RS",           "DT",           "UC",           "PD",
  65         "Sp",           "Vb",           "Ve",
  66         };
  67
  68 const   char * const *man_macronames = __man_macronames;
  69
  70 static  struct man_node *man_node_alloc(int, int,
  71                                 enum man_type, enum mant);
  72 static  int              man_node_append(struct man *,
  73                                 struct man_node *);
  74 static  void             man_node_free(struct man_node *);
  75 static  void             man_node_unlink(struct man *,
  76                                 struct man_node *);
  77 static  int              man_ptext(struct man *, int, char *);
  78 static  int              man_pmacro(struct man *, int, char *);
  79 static  void             man_free1(struct man *);
  80 static  void             man_alloc1(struct man *);
  81 static  int              macrowarn(struct man *, int, const char *);
  82
  83
  84 const struct man_node *
  85 man_node(const struct man *m)
  86 {
  87
  88         return(MAN_HALT & m->flags ? NULL : m->first);
  89 }
  90
  91
  92 const struct man_meta *
  93 man_meta(const struct man *m)
  94 {
  95
  96         return(MAN_HALT & m->flags ? NULL : &m->meta);
  97 }
  98
  99
 100 void
 101 man_reset(struct man *man)
 102 {
 103
 104         man_free1(man);
 105         man_alloc1(man);
 106 }
 107
 108
 109 void
 110 man_free(struct man *man)
 111 {
 112
 113         man_free1(man);
 114         free(man);
 115 }
 116
 117
 118 struct man *
 119 man_alloc(void *data, int pflags, const struct man_cb *cb)
 120 {
 121         struct man      *p;
 122
 123         p = mandoc_calloc(1, sizeof(struct man));
 124
 125         if (cb)
 126                 memcpy(&p->cb, cb, sizeof(struct man_cb));
 127
 128         man_hash_init();
 129         p->data = data;
 130         p->pflags = pflags;
 131
 132         man_alloc1(p);
 133         return(p);
 134 }
 135
 136
 137 int
 138 man_endparse(struct man *m)
 139 {
 140
 141         if (MAN_HALT & m->flags)
 142                 return(0);
 143         else if (man_macroend(m))
 144                 return(1);
 145         m->flags |= MAN_HALT;
 146         return(0);
 147 }
 148
 149
 150 int
 151 man_parseln(struct man *m, int ln, char *buf)
 152 {
 153
 154         return(('.' == *buf || '\'' == *buf) ?
 155                         man_pmacro(m, ln, buf) :
 156                         man_ptext(m, ln, buf));
 157 }
 158
 159
 160 static void
 161 man_free1(struct man *man)
 162 {
 163
 164         if (man->first)
 165                 man_node_delete(man, man->first);
 166         if (man->meta.title)
 167                 free(man->meta.title);
 168         if (man->meta.source)
 169                 free(man->meta.source);
 170         if (man->meta.vol)
 171                 free(man->meta.vol);
 172         if (man->meta.msec)
 173                 free(man->meta.msec);
 174 }
 175
 176
 177 static void
 178 man_alloc1(struct man *m)
 179 {
 180
 181         memset(&m->meta, 0, sizeof(struct man_meta));
 182         m->flags = 0;
 183         m->last = mandoc_calloc(1, sizeof(struct man_node));
 184         m->first = m->last;
 185         m->last->type = MAN_ROOT;
 186         m->last->tok = MAN_MAX;
 187         m->next = MAN_NEXT_CHILD;
 188 }
 189
 190
 191 static int
 192 man_node_append(struct man *man, struct man_node *p)
 193 {
 194
 195         assert(man->last);
 196         assert(man->first);
 197         assert(MAN_ROOT != p->type);
 198
 199         switch (man->next) {
 200         case (MAN_NEXT_SIBLING):
 201                 man->last->next = p;
 202                 p->prev = man->last;
 203                 p->parent = man->last->parent;
 204                 break;
 205         case (MAN_NEXT_CHILD):
 206                 man->last->child = p;
 207                 p->parent = man->last;
 208                 break;
 209         default:
 210                 abort();
 211                 /* NOTREACHED */
 212         }
 213
 214         assert(p->parent);
 215         p->parent->nchild++;
 216
 217         if ( ! man_valid_pre(man, p))
 218                 return(0);
 219
 220         switch (p->type) {
 221         case (MAN_HEAD):
 222                 assert(MAN_BLOCK == p->parent->type);
 223                 p->parent->head = p;
 224                 break;
 225         case (MAN_BODY):
 226                 assert(MAN_BLOCK == p->parent->type);
 227                 p->parent->body = p;
 228                 break;
 229         default:
 230                 break;
 231         }
 232
 233         man->last = p;
 234
 235         switch (p->type) {
 236         case (MAN_TEXT):
 237                 if ( ! man_valid_post(man))
 238                         return(0);
 239                 if ( ! man_action_post(man))
 240                         return(0);
 241                 break;
 242         default:
 243                 break;
 244         }
 245
 246         return(1);
 247 }
 248
 249
 250 static struct man_node *
 251 man_node_alloc(int line, int pos, enum man_type type, enum mant tok)
 252 {
 253         struct man_node *p;
 254
 255         p = mandoc_calloc(1, sizeof(struct man_node));
 256         p->line = line;
 257         p->pos = pos;
 258         p->type = type;
 259         p->tok = tok;
 260         return(p);
 261 }
 262
 263
 264 int
 265 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 266 {
 267         struct man_node *p;
 268
 269         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 270         if ( ! man_node_append(m, p))
 271                 return(0);
 272         m->next = MAN_NEXT_CHILD;
 273         return(1);
 274 }
 275
 276
 277 int
 278 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 279 {
 280         struct man_node *p;
 281
 282         p = man_node_alloc(line, pos, MAN_HEAD, tok);
 283         if ( ! man_node_append(m, p))
 284                 return(0);
 285         m->next = MAN_NEXT_CHILD;
 286         return(1);
 287 }
 288
 289
 290 int
 291 man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 292 {
 293         struct man_node *p;
 294
 295         p = man_node_alloc(line, pos, MAN_BODY, tok);
 296         if ( ! man_node_append(m, p))
 297                 return(0);
 298         m->next = MAN_NEXT_CHILD;
 299         return(1);
 300 }
 301
 302
 303 int
 304 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 305 {
 306         struct man_node *p;
 307
 308         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 309         if ( ! man_node_append(m, p))
 310                 return(0);
 311         m->next = MAN_NEXT_CHILD;
 312         return(1);
 313 }
 314
 315
 316 int
 317 man_word_alloc(struct man *m, int line, int pos, const char *word)
 318 {
 319         struct man_node *n;
 320         size_t           sv, len;
 321
 322         len = strlen(word);
 323
 324         n = man_node_alloc(line, pos, MAN_TEXT, MAN_MAX);
 325         n->string = mandoc_malloc(len + 1);
 326         sv = strlcpy(n->string, word, len + 1);
 327
 328         /* Prohibit truncation. */
 329         assert(sv < len + 1);
 330
 331         if ( ! man_node_append(m, n))
 332                 return(0);
 333
 334         m->next = MAN_NEXT_SIBLING;
 335         return(1);
 336 }
 337
 338
 339 /*
 340  * Free all of the resources held by a node.  This does NOT unlink a
 341  * node from its context; for that, see man_node_unlink().
 342  */
 343 static void
 344 man_node_free(struct man_node *p)
 345 {
 346
 347         if (p->string)
 348                 free(p->string);
 349         free(p);
 350 }
 351
 352
 353 void
 354 man_node_delete(struct man *m, struct man_node *p)
 355 {
 356
 357         while (p->child)
 358                 man_node_delete(m, p->child);
 359
 360         man_node_unlink(m, p);
 361         man_node_free(p);
 362 }
 363
 364
 365 static int
 366 man_ptext(struct man *m, int line, char *buf)
 367 {
 368         int              i;
 369
 370         /* Ignore bogus comments. */
 371
 372         if ('\\' == buf[0] && '.' == buf[1] && '\"' == buf[2])
 373                 return(man_pwarn(m, line, 0, WBADCOMMENT));
 374
 375         /* Literal free-form text whitespace is preserved. */
 376
 377         if (MAN_LITERAL & m->flags) {
 378                 if ( ! man_word_alloc(m, line, 0, buf))
 379                         return(0);
 380                 goto descope;
 381         }
 382
 383         /* Pump blank lines directly into the backend. */
 384
 385         for (i = 0; ' ' == buf[i]; i++)
 386                 /* Skip leading whitespace. */ ;
 387
 388         if ('\0' == buf[i]) {
 389                 /* Allocate a blank entry. */
 390                 if ( ! man_word_alloc(m, line, 0, ""))
 391                         return(0);
 392                 goto descope;
 393         }
 394
 395         /*
 396          * Warn if the last un-escaped character is whitespace. Then
 397          * strip away the remaining spaces (tabs stay!).
 398          */
 399
 400         i = (int)strlen(buf);
 401         assert(i);
 402
 403         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 404                 if (i > 1 && '\\' != buf[i - 2])
 405                         if ( ! man_pwarn(m, line, i - 1, WTSPACE))
 406                                 return(0);
 407
 408                 for (--i; i && ' ' == buf[i]; i--)
 409                         /* Spin back to non-space. */ ;
 410
 411                 /* Jump ahead of escaped whitespace. */
 412                 i += '\\' == buf[i] ? 2 : 1;
 413
 414                 buf[i] = '\0';
 415         }
 416
 417         if ( ! man_word_alloc(m, line, 0, buf))
 418                 return(0);
 419
 420         /*
 421          * End-of-sentence check.  If the last character is an unescaped
 422          * EOS character, then flag the node as being the end of a
 423          * sentence.  The front-end will know how to interpret this.
 424          */
 425
 426         /* FIXME: chain of close delims. */
 427
 428         assert(i);
 429
 430         if (mandoc_eos(buf, (size_t)i))
 431                 m->last->flags |= MAN_EOS;
 432
 433 descope:
 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, WERRMAX))
 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, WERRMAX))
 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, "unknown macro: %s%s",
 461                                 buf, strlen(buf) > 3 ? "..." : ""));
 462         return(man_vwarn(m, ln, 0, "unknown macro: %s%s",
 463                                 buf, strlen(buf) > 3 ? "..." : ""));
 464 }
 465
 466
 467 int
 468 man_pmacro(struct man *m, int ln, char *buf)
 469 {
 470         int              i, j, ppos;
 471         enum mant        tok;
 472         char             mac[5];
 473         struct man_node *n;
 474
 475         /* Comments and empties are quickly ignored. */
 476
 477         if ('\0' == buf[1])
 478                 return(1);
 479
 480         i = 1;
 481
 482         /*
 483          * Skip whitespace between the control character and initial
 484          * text.  "Whitespace" is both spaces and tabs.
 485          */
 486
 487         if (' ' == buf[i] || '\t' == buf[i]) {
 488                 i++;
 489                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 490                         i++;
 491                 if ('\0' == buf[i])
 492                         goto out;
 493         }
 494
 495         ppos = i;
 496
 497         /* Copy the first word into a nil-terminated buffer. */
 498
 499         for (j = 0; j < 4; j++, i++) {
 500                 if ('\0' == (mac[j] = buf[i]))
 501                         break;
 502                 else if (' ' == buf[i])
 503                         break;
 504
 505                 /* Check for invalid characters. */
 506
 507                 if (isgraph((u_char)buf[i]))
 508                         continue;
 509                 return(man_perr(m, ln, i, WNPRINT));
 510         }
 511
 512         mac[j] = '\0';
 513
 514         if (j == 4 || j < 1) {
 515                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 516                         (void)man_perr(m, ln, ppos, WMACROFORM);
 517                         goto err;
 518                 }
 519                 if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
 520                         goto err;
 521                 return(1);
 522         }
 523
 524         if (MAN_MAX == (tok = man_hash_find(mac))) {
 525                 if ( ! macrowarn(m, ln, mac))
 526                         goto err;
 527                 return(1);
 528         }
 529
 530         /* The macro is sane.  Jump to the next word. */
 531
 532         while (buf[i] && ' ' == buf[i])
 533                 i++;
 534
 535         /*
 536          * Trailing whitespace.  Note that tabs are allowed to be passed
 537          * into the parser as "text", so we only warn about spaces here.
 538          */
 539
 540         if ('\0' == buf[i] && ' ' == buf[i - 1])
 541                 if ( ! man_pwarn(m, ln, i - 1, WTSPACE))
 542                         goto err;
 543
 544         /*
 545          * Remove prior ELINE macro, as it's being clobbering by a new
 546          * macro.  Note that NSCOPED macros do not close out ELINE
 547          * macros---they don't print text---so we let those slip by.
 548          */
 549
 550         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 551                         m->flags & MAN_ELINE) {
 552                 assert(MAN_TEXT != m->last->type);
 553
 554                 /*
 555                  * This occurs in the following construction:
 556                  *   .B
 557                  *   .br
 558                  *   .B
 559                  *   .br
 560                  *   I hate man macros.
 561                  * Flat-out disallow this madness.
 562                  */
 563                 if (MAN_NSCOPED & man_macros[m->last->tok].flags)
 564                         return(man_perr(m, ln, ppos, WLNSCOPE));
 565
 566                 n = m->last;
 567
 568                 assert(n);
 569                 assert(NULL == n->child);
 570                 assert(0 == n->nchild);
 571
 572                 if ( ! man_nwarn(m, n, WLNSCOPE))
 573                         return(0);
 574
 575                 man_node_delete(m, n);
 576                 m->flags &= ~MAN_ELINE;
 577         }
 578
 579         /*
 580          * Save the fact that we're in the next-line for a block.  In
 581          * this way, embedded roff instructions can "remember" state
 582          * when they exit.
 583          */
 584
 585         if (MAN_BLINE & m->flags)
 586                 m->flags |= MAN_BPLINE;
 587
 588         /* Call to handler... */
 589
 590         assert(man_macros[tok].fp);
 591         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 592                 goto err;
 593
 594 out:
 595         /*
 596          * We weren't in a block-line scope when entering the
 597          * above-parsed macro, so return.
 598          */
 599
 600         if ( ! (MAN_BPLINE & m->flags)) {
 601                 m->flags &= ~MAN_ILINE;
 602                 return(1);
 603         }
 604         m->flags &= ~MAN_BPLINE;
 605
 606         /*
 607          * If we're in a block scope, then allow this macro to slip by
 608          * without closing scope around it.
 609          */
 610
 611         if (MAN_ILINE & m->flags) {
 612                 m->flags &= ~MAN_ILINE;
 613                 return(1);
 614         }
 615
 616         /*
 617          * If we've opened a new next-line element scope, then return
 618          * now, as the next line will close out the block scope.
 619          */
 620
 621         if (MAN_ELINE & m->flags)
 622                 return(1);
 623
 624         /* Close out the block scope opened in the prior line.  */
 625
 626         assert(MAN_BLINE & m->flags);
 627         m->flags &= ~MAN_BLINE;
 628
 629         if ( ! man_unscope(m, m->last->parent, WERRMAX))
 630                 return(0);
 631         return(man_body_alloc(m, ln, 0, m->last->tok));
 632
 633 err:    /* Error out. */
 634
 635         m->flags |= MAN_HALT;
 636         return(0);
 637 }
 638
 639
 640 int
 641 man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
 642 {
 643         char             buf[256];
 644         va_list          ap;
 645
 646         if (NULL == man->cb.man_err)
 647                 return(0);
 648
 649         va_start(ap, fmt);
 650         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 651         va_end(ap);
 652         return((*man->cb.man_err)(man->data, ln, pos, buf));
 653 }
 654
 655
 656 int
 657 man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 658 {
 659         char             buf[256];
 660         va_list          ap;
 661
 662         if (NULL == man->cb.man_warn)
 663                 return(0);
 664
 665         va_start(ap, fmt);
 666         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 667         va_end(ap);
 668         return((*man->cb.man_warn)(man->data, ln, pos, buf));
 669 }
 670
 671
 672 int
 673 man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 674 {
 675         const char       *p;
 676
 677         p = __man_merrnames[(int)type];
 678         assert(p);
 679
 680         if (iserr)
 681                 return(man_verr(m, line, pos, p));
 682
 683         return(man_vwarn(m, line, pos, p));
 684 }
 685
 686
 687 /*
 688  * Unlink a node from its context.  If "m" is provided, the last parse
 689  * point will also be adjusted accordingly.
 690  */
 691 static void
 692 man_node_unlink(struct man *m, struct man_node *n)
 693 {
 694
 695         /* Adjust siblings. */
 696
 697         if (n->prev)
 698                 n->prev->next = n->next;
 699         if (n->next)
 700                 n->next->prev = n->prev;
 701
 702         /* Adjust parent. */
 703
 704         if (n->parent) {
 705                 n->parent->nchild--;
 706                 if (n->parent->child == n)
 707                         n->parent->child = n->prev ? n->prev : n->next;
 708         }
 709
 710         /* Adjust parse point, if applicable. */
 711
 712         if (m && m->last == n) {
 713                 /*XXX: this can occur when bailing from validation. */
 714                 /*assert(NULL == n->next);*/
 715                 if (n->prev) {
 716                         m->last = n->prev;
 717                         m->next = MAN_NEXT_SIBLING;
 718                 } else {
 719                         m->last = n->parent;
 720                         m->next = MAN_NEXT_CHILD;
 721                 }
 722         }
 723
 724         if (m && m->first == n)
 725                 m->first = NULL;
 726 }