man.c

   1 /*      $Id: man.c,v 1.72 2010/05/16 10:59:36 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 *, int);
  78 static  int              man_pmacro(struct man *, int, char *, int);
  79 static  void             man_free1(struct man *);
  80 static  void             man_alloc1(struct man *);
  81 static  int              macrowarn(struct man *, int, const char *, int);
  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, int offs)
 152 {
 153
 154         if (MAN_HALT & m->flags)
 155                 return(0);
 156
 157         return(('.' == buf[offs] || '\'' == buf[offs]) ?
 158                         man_pmacro(m, ln, buf, offs) :
 159                         man_ptext(m, ln, buf, offs));
 160 }
 161
 162
 163 static void
 164 man_free1(struct man *man)
 165 {
 166
 167         if (man->first)
 168                 man_node_delete(man, man->first);
 169         if (man->meta.title)
 170                 free(man->meta.title);
 171         if (man->meta.source)
 172                 free(man->meta.source);
 173         if (man->meta.vol)
 174                 free(man->meta.vol);
 175         if (man->meta.msec)
 176                 free(man->meta.msec);
 177 }
 178
 179
 180 static void
 181 man_alloc1(struct man *m)
 182 {
 183
 184         memset(&m->meta, 0, sizeof(struct man_meta));
 185         m->flags = 0;
 186         m->last = mandoc_calloc(1, sizeof(struct man_node));
 187         m->first = m->last;
 188         m->last->type = MAN_ROOT;
 189         m->last->tok = MAN_MAX;
 190         m->next = MAN_NEXT_CHILD;
 191 }
 192
 193
 194 static int
 195 man_node_append(struct man *man, struct man_node *p)
 196 {
 197
 198         assert(man->last);
 199         assert(man->first);
 200         assert(MAN_ROOT != p->type);
 201
 202         switch (man->next) {
 203         case (MAN_NEXT_SIBLING):
 204                 man->last->next = p;
 205                 p->prev = man->last;
 206                 p->parent = man->last->parent;
 207                 break;
 208         case (MAN_NEXT_CHILD):
 209                 man->last->child = p;
 210                 p->parent = man->last;
 211                 break;
 212         default:
 213                 abort();
 214                 /* NOTREACHED */
 215         }
 216
 217         assert(p->parent);
 218         p->parent->nchild++;
 219
 220         if ( ! man_valid_pre(man, p))
 221                 return(0);
 222
 223         switch (p->type) {
 224         case (MAN_HEAD):
 225                 assert(MAN_BLOCK == p->parent->type);
 226                 p->parent->head = p;
 227                 break;
 228         case (MAN_BODY):
 229                 assert(MAN_BLOCK == p->parent->type);
 230                 p->parent->body = p;
 231                 break;
 232         default:
 233                 break;
 234         }
 235
 236         man->last = p;
 237
 238         switch (p->type) {
 239         case (MAN_TEXT):
 240                 if ( ! man_valid_post(man))
 241                         return(0);
 242                 if ( ! man_action_post(man))
 243                         return(0);
 244                 break;
 245         default:
 246                 break;
 247         }
 248
 249         return(1);
 250 }
 251
 252
 253 static struct man_node *
 254 man_node_alloc(int line, int pos, enum man_type type, enum mant tok)
 255 {
 256         struct man_node *p;
 257
 258         p = mandoc_calloc(1, sizeof(struct man_node));
 259         p->line = line;
 260         p->pos = pos;
 261         p->type = type;
 262         p->tok = tok;
 263         return(p);
 264 }
 265
 266
 267 int
 268 man_elem_alloc(struct man *m, int line, int pos, enum mant tok)
 269 {
 270         struct man_node *p;
 271
 272         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 273         if ( ! man_node_append(m, p))
 274                 return(0);
 275         m->next = MAN_NEXT_CHILD;
 276         return(1);
 277 }
 278
 279
 280 int
 281 man_head_alloc(struct man *m, int line, int pos, enum mant tok)
 282 {
 283         struct man_node *p;
 284
 285         p = man_node_alloc(line, pos, MAN_HEAD, tok);
 286         if ( ! man_node_append(m, p))
 287                 return(0);
 288         m->next = MAN_NEXT_CHILD;
 289         return(1);
 290 }
 291
 292
 293 int
 294 man_body_alloc(struct man *m, int line, int pos, enum mant tok)
 295 {
 296         struct man_node *p;
 297
 298         p = man_node_alloc(line, pos, MAN_BODY, tok);
 299         if ( ! man_node_append(m, p))
 300                 return(0);
 301         m->next = MAN_NEXT_CHILD;
 302         return(1);
 303 }
 304
 305
 306 int
 307 man_block_alloc(struct man *m, int line, int pos, enum mant tok)
 308 {
 309         struct man_node *p;
 310
 311         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 312         if ( ! man_node_append(m, p))
 313                 return(0);
 314         m->next = MAN_NEXT_CHILD;
 315         return(1);
 316 }
 317
 318
 319 int
 320 man_word_alloc(struct man *m, int line, int pos, const char *word)
 321 {
 322         struct man_node *n;
 323         size_t           sv, len;
 324
 325         len = strlen(word);
 326
 327         n = man_node_alloc(line, pos, MAN_TEXT, MAN_MAX);
 328         n->string = mandoc_malloc(len + 1);
 329         sv = strlcpy(n->string, word, len + 1);
 330
 331         /* Prohibit truncation. */
 332         assert(sv < len + 1);
 333
 334         if ( ! man_node_append(m, n))
 335                 return(0);
 336
 337         m->next = MAN_NEXT_SIBLING;
 338         return(1);
 339 }
 340
 341
 342 /*
 343  * Free all of the resources held by a node.  This does NOT unlink a
 344  * node from its context; for that, see man_node_unlink().
 345  */
 346 static void
 347 man_node_free(struct man_node *p)
 348 {
 349
 350         if (p->string)
 351                 free(p->string);
 352         free(p);
 353 }
 354
 355
 356 void
 357 man_node_delete(struct man *m, struct man_node *p)
 358 {
 359
 360         while (p->child)
 361                 man_node_delete(m, p->child);
 362
 363         man_node_unlink(m, p);
 364         man_node_free(p);
 365 }
 366
 367
 368 static int
 369 man_ptext(struct man *m, int line, char *buf, int offs)
 370 {
 371         int              i;
 372
 373         /* Ignore bogus comments. */
 374
 375         if ('\\' == buf[offs] &&
 376                         '.' == buf[offs + 1] &&
 377                         '"' == buf[offs + 2])
 378                 return(man_pwarn(m, line, offs, WBADCOMMENT));
 379
 380         /* Literal free-form text whitespace is preserved. */
 381
 382         if (MAN_LITERAL & m->flags) {
 383                 if ( ! man_word_alloc(m, line, offs, buf + offs))
 384                         return(0);
 385                 goto descope;
 386         }
 387
 388         /* Pump blank lines directly into the backend. */
 389
 390         for (i = offs; ' ' == buf[i]; i++)
 391                 /* Skip leading whitespace. */ ;
 392
 393         if ('\0' == buf[i]) {
 394                 /* Allocate a blank entry. */
 395                 if ( ! man_word_alloc(m, line, offs, ""))
 396                         return(0);
 397                 goto descope;
 398         }
 399
 400         /*
 401          * Warn if the last un-escaped character is whitespace. Then
 402          * strip away the remaining spaces (tabs stay!).
 403          */
 404
 405         i = (int)strlen(buf);
 406         assert(i);
 407
 408         if (' ' == buf[i - 1] || '\t' == buf[i - 1]) {
 409                 if (i > 1 && '\\' != buf[i - 2])
 410                         if ( ! man_pwarn(m, line, i - 1, WTSPACE))
 411                                 return(0);
 412
 413                 for (--i; i && ' ' == buf[i]; i--)
 414                         /* Spin back to non-space. */ ;
 415
 416                 /* Jump ahead of escaped whitespace. */
 417                 i += '\\' == buf[i] ? 2 : 1;
 418
 419                 buf[i] = '\0';
 420         }
 421
 422         if ( ! man_word_alloc(m, line, offs, buf + offs))
 423                 return(0);
 424
 425         /*
 426          * End-of-sentence check.  If the last character is an unescaped
 427          * EOS character, then flag the node as being the end of a
 428          * sentence.  The front-end will know how to interpret this.
 429          */
 430
 431         assert(i);
 432         if (mandoc_eos(buf, (size_t)i))
 433                 m->last->flags |= MAN_EOS;
 434
 435 descope:
 436         /*
 437          * Co-ordinate what happens with having a next-line scope open:
 438          * first close out the element scope (if applicable), then close
 439          * out the block scope (also if applicable).
 440          */
 441
 442         if (MAN_ELINE & m->flags) {
 443                 m->flags &= ~MAN_ELINE;
 444                 if ( ! man_unscope(m, m->last->parent, WERRMAX))
 445                         return(0);
 446         }
 447
 448         if ( ! (MAN_BLINE & m->flags))
 449                 return(1);
 450         m->flags &= ~MAN_BLINE;
 451
 452         if ( ! man_unscope(m, m->last->parent, WERRMAX))
 453                 return(0);
 454         return(man_body_alloc(m, line, offs, m->last->tok));
 455 }
 456
 457
 458 static int
 459 macrowarn(struct man *m, int ln, const char *buf, int offs)
 460 {
 461         if ( ! (MAN_IGN_MACRO & m->pflags))
 462                 return(man_verr(m, ln, offs, "unknown macro: %s%s",
 463                                 buf, strlen(buf) > 3 ? "..." : ""));
 464         return(man_vwarn(m, ln, offs, "unknown macro: %s%s",
 465                                 buf, strlen(buf) > 3 ? "..." : ""));
 466 }
 467
 468
 469 int
 470 man_pmacro(struct man *m, int ln, char *buf, int offs)
 471 {
 472         int              i, j, ppos;
 473         enum mant        tok;
 474         char             mac[5];
 475         struct man_node *n;
 476
 477         /* Comments and empties are quickly ignored. */
 478
 479         offs++;
 480
 481         if ('\0' == buf[offs])
 482                 return(1);
 483
 484         i = offs;
 485
 486         /*
 487          * Skip whitespace between the control character and initial
 488          * text.  "Whitespace" is both spaces and tabs.
 489          */
 490
 491         if (' ' == buf[i] || '\t' == buf[i]) {
 492                 i++;
 493                 while (buf[i] && (' ' == buf[i] || '\t' == buf[i]))
 494                         i++;
 495                 if ('\0' == buf[i])
 496                         goto out;
 497         }
 498
 499         ppos = i;
 500
 501         /* Copy the first word into a nil-terminated buffer. */
 502
 503         for (j = 0; j < 4; j++, i++) {
 504                 if ('\0' == (mac[j] = buf[i]))
 505                         break;
 506                 else if (' ' == buf[i])
 507                         break;
 508
 509                 /* Check for invalid characters. */
 510
 511                 if (isgraph((u_char)buf[i]))
 512                         continue;
 513                 return(man_perr(m, ln, i, WNPRINT));
 514         }
 515
 516         mac[j] = '\0';
 517
 518         if (j == 4 || j < 1) {
 519                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 520                         (void)man_perr(m, ln, ppos, WMACROFORM);
 521                         goto err;
 522                 }
 523                 if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
 524                         goto err;
 525                 return(1);
 526         }
 527
 528         if (MAN_MAX == (tok = man_hash_find(mac))) {
 529                 if ( ! macrowarn(m, ln, mac, ppos))
 530                         goto err;
 531                 return(1);
 532         }
 533
 534         /* The macro is sane.  Jump to the next word. */
 535
 536         while (buf[i] && ' ' == buf[i])
 537                 i++;
 538
 539         /*
 540          * Trailing whitespace.  Note that tabs are allowed to be passed
 541          * into the parser as "text", so we only warn about spaces here.
 542          */
 543
 544         if ('\0' == buf[i] && ' ' == buf[i - 1])
 545                 if ( ! man_pwarn(m, ln, i - 1, WTSPACE))
 546                         goto err;
 547
 548         /*
 549          * Remove prior ELINE macro, as it's being clobbering by a new
 550          * macro.  Note that NSCOPED macros do not close out ELINE
 551          * macros---they don't print text---so we let those slip by.
 552          */
 553
 554         if ( ! (MAN_NSCOPED & man_macros[tok].flags) &&
 555                         m->flags & MAN_ELINE) {
 556                 assert(MAN_TEXT != m->last->type);
 557
 558                 /*
 559                  * This occurs in the following construction:
 560                  *   .B
 561                  *   .br
 562                  *   .B
 563                  *   .br
 564                  *   I hate man macros.
 565                  * Flat-out disallow this madness.
 566                  */
 567                 if (MAN_NSCOPED & man_macros[m->last->tok].flags)
 568                         return(man_perr(m, ln, ppos, WLNSCOPE));
 569
 570                 n = m->last;
 571
 572                 assert(n);
 573                 assert(NULL == n->child);
 574                 assert(0 == n->nchild);
 575
 576                 if ( ! man_nwarn(m, n, WLNSCOPE))
 577                         return(0);
 578
 579                 man_node_delete(m, n);
 580                 m->flags &= ~MAN_ELINE;
 581         }
 582
 583         /*
 584          * Save the fact that we're in the next-line for a block.  In
 585          * this way, embedded roff instructions can "remember" state
 586          * when they exit.
 587          */
 588
 589         if (MAN_BLINE & m->flags)
 590                 m->flags |= MAN_BPLINE;
 591
 592         /* Call to handler... */
 593
 594         assert(man_macros[tok].fp);
 595         if ( ! (*man_macros[tok].fp)(m, tok, ln, ppos, &i, buf))
 596                 goto err;
 597
 598 out:
 599         /*
 600          * We weren't in a block-line scope when entering the
 601          * above-parsed macro, so return.
 602          */
 603
 604         if ( ! (MAN_BPLINE & m->flags)) {
 605                 m->flags &= ~MAN_ILINE;
 606                 return(1);
 607         }
 608         m->flags &= ~MAN_BPLINE;
 609
 610         /*
 611          * If we're in a block scope, then allow this macro to slip by
 612          * without closing scope around it.
 613          */
 614
 615         if (MAN_ILINE & m->flags) {
 616                 m->flags &= ~MAN_ILINE;
 617                 return(1);
 618         }
 619
 620         /*
 621          * If we've opened a new next-line element scope, then return
 622          * now, as the next line will close out the block scope.
 623          */
 624
 625         if (MAN_ELINE & m->flags)
 626                 return(1);
 627
 628         /* Close out the block scope opened in the prior line.  */
 629
 630         assert(MAN_BLINE & m->flags);
 631         m->flags &= ~MAN_BLINE;
 632
 633         if ( ! man_unscope(m, m->last->parent, WERRMAX))
 634                 return(0);
 635         return(man_body_alloc(m, ln, offs, m->last->tok));
 636
 637 err:    /* Error out. */
 638
 639         m->flags |= MAN_HALT;
 640         return(0);
 641 }
 642
 643
 644 int
 645 man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
 646 {
 647         char             buf[256];
 648         va_list          ap;
 649
 650         if (NULL == man->cb.man_err)
 651                 return(0);
 652
 653         va_start(ap, fmt);
 654         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 655         va_end(ap);
 656         return((*man->cb.man_err)(man->data, ln, pos, buf));
 657 }
 658
 659
 660 int
 661 man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 662 {
 663         char             buf[256];
 664         va_list          ap;
 665
 666         if (NULL == man->cb.man_warn)
 667                 return(0);
 668
 669         va_start(ap, fmt);
 670         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 671         va_end(ap);
 672         return((*man->cb.man_warn)(man->data, ln, pos, buf));
 673 }
 674
 675
 676 int
 677 man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 678 {
 679         const char       *p;
 680
 681         p = __man_merrnames[(int)type];
 682         assert(p);
 683
 684         if (iserr)
 685                 return(man_verr(m, line, pos, p));
 686
 687         return(man_vwarn(m, line, pos, p));
 688 }
 689
 690
 691 /*
 692  * Unlink a node from its context.  If "m" is provided, the last parse
 693  * point will also be adjusted accordingly.
 694  */
 695 static void
 696 man_node_unlink(struct man *m, struct man_node *n)
 697 {
 698
 699         /* Adjust siblings. */
 700
 701         if (n->prev)
 702                 n->prev->next = n->next;
 703         if (n->next)
 704                 n->next->prev = n->prev;
 705
 706         /* Adjust parent. */
 707
 708         if (n->parent) {
 709                 n->parent->nchild--;
 710                 if (n->parent->child == n)
 711                         n->parent->child = n->prev ? n->prev : n->next;
 712         }
 713
 714         /* Adjust parse point, if applicable. */
 715
 716         if (m && m->last == n) {
 717                 /*XXX: this can occur when bailing from validation. */
 718                 /*assert(NULL == n->next);*/
 719                 if (n->prev) {
 720                         m->last = n->prev;
 721                         m->next = MAN_NEXT_SIBLING;
 722                 } else {
 723                         m->last = n->parent;
 724                         m->next = MAN_NEXT_CHILD;
 725                 }
 726         }
 727
 728         if (m && m->first == n)
 729                 m->first = NULL;
 730 }