usr.bin/mandoc/man.c

   1 /*      $Id: man.c,v 1.49 2010/01/07 10:24:43 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 #include <sys/types.h>
  18
  19 #include <assert.h>
  20 #include <ctype.h>
  21 #include <stdarg.h>
  22 #include <stdlib.h>
  23 #include <stdio.h>
  24 #include <string.h>
  25
  26 #include "libman.h"
  27 #include "libmandoc.h"
  28
  29 const   char *const __man_merrnames[WERRMAX] = {
  30         "invalid character", /* WNPRINT */
  31         "invalid manual section", /* WMSEC */
  32         "invalid date format", /* WDATE */
  33         "scope of prior line violated", /* WLNSCOPE */
  34         "trailing whitespace", /* WTSPACE */
  35         "unterminated quoted parameter", /* WTQUOTE */
  36         "document has no body", /* WNODATA */
  37         "document has no title/section", /* WNOTITLE */
  38         "invalid escape sequence", /* WESCAPE */
  39         "invalid number format", /* WNUMFMT */
  40         "expected block head arguments", /* WHEADARGS */
  41         "expected block body arguments", /* WBODYARGS */
  42         "expected empty block head", /* WNHEADARGS */
  43         "ill-formed macro", /* WMACROFORM */
  44         "scope open on exit", /* WEXITSCOPE */
  45         "no scope context", /* WNOSCOPE */
  46         "literal context already open", /* WOLITERAL */
  47         "no literal context open" /* WNLITERAL */
  48 };
  49
  50 const   char *const __man_macronames[MAN_MAX] = {
  51         "br",           "TH",           "SH",           "SS",
  52         "TP",           "LP",           "PP",           "P",
  53         "IP",           "HP",           "SM",           "SB",
  54         "BI",           "IB",           "BR",           "RB",
  55         "R",            "B",            "I",            "IR",
  56         "RI",           "na",           "i",            "sp",
  57         "nf",           "fi",           "r",            "RE",
  58         "RS",           "DT",           "UC",           "PD"
  59         };
  60
  61 const   char * const *man_macronames = __man_macronames;
  62
  63 static  struct man_node *man_node_alloc(int, int,
  64                                 enum man_type, int);
  65 static  int              man_node_append(struct man *,
  66                                 struct man_node *);
  67 static  int              man_ptext(struct man *, int, char *);
  68 static  int              man_pmacro(struct man *, int, char *);
  69 static  void             man_free1(struct man *);
  70 static  void             man_alloc1(struct man *);
  71 static  int              pstring(struct man *, int, int,
  72                                 const char *, size_t);
  73 static  int              macrowarn(struct man *, int, const char *);
  74
  75
  76 const struct man_node *
  77 man_node(const struct man *m)
  78 {
  79
  80         return(MAN_HALT & m->flags ? NULL : m->first);
  81 }
  82
  83
  84 const struct man_meta *
  85 man_meta(const struct man *m)
  86 {
  87
  88         return(MAN_HALT & m->flags ? NULL : &m->meta);
  89 }
  90
  91
  92 void
  93 man_reset(struct man *man)
  94 {
  95
  96         man_free1(man);
  97         man_alloc1(man);
  98 }
  99
 100
 101 void
 102 man_free(struct man *man)
 103 {
 104
 105         man_free1(man);
 106         free(man);
 107 }
 108
 109
 110 struct man *
 111 man_alloc(void *data, int pflags, const struct man_cb *cb)
 112 {
 113         struct man      *p;
 114
 115         p = mandoc_calloc(1, sizeof(struct man));
 116
 117         if (cb)
 118                 memcpy(&p->cb, cb, sizeof(struct man_cb));
 119
 120         man_hash_init();
 121         p->data = data;
 122         p->pflags = pflags;
 123
 124         man_alloc1(p);
 125         return(p);
 126 }
 127
 128
 129 int
 130 man_endparse(struct man *m)
 131 {
 132
 133         if (MAN_HALT & m->flags)
 134                 return(0);
 135         else if (man_macroend(m))
 136                 return(1);
 137         m->flags |= MAN_HALT;
 138         return(0);
 139 }
 140
 141
 142 int
 143 man_parseln(struct man *m, int ln, char *buf)
 144 {
 145
 146         return('.' == *buf ?
 147                         man_pmacro(m, ln, buf) :
 148                         man_ptext(m, ln, buf));
 149 }
 150
 151
 152 static void
 153 man_free1(struct man *man)
 154 {
 155
 156         if (man->first)
 157                 man_node_freelist(man->first);
 158         if (man->meta.title)
 159                 free(man->meta.title);
 160         if (man->meta.source)
 161                 free(man->meta.source);
 162         if (man->meta.vol)
 163                 free(man->meta.vol);
 164 }
 165
 166
 167 static void
 168 man_alloc1(struct man *m)
 169 {
 170
 171         memset(&m->meta, 0, sizeof(struct man_meta));
 172         m->flags = 0;
 173         m->last = mandoc_calloc(1, sizeof(struct man_node));
 174         m->first = m->last;
 175         m->last->type = MAN_ROOT;
 176         m->next = MAN_NEXT_CHILD;
 177 }
 178
 179
 180 static int
 181 man_node_append(struct man *man, struct man_node *p)
 182 {
 183
 184         assert(man->last);
 185         assert(man->first);
 186         assert(MAN_ROOT != p->type);
 187
 188         switch (man->next) {
 189         case (MAN_NEXT_SIBLING):
 190                 man->last->next = p;
 191                 p->prev = man->last;
 192                 p->parent = man->last->parent;
 193                 break;
 194         case (MAN_NEXT_CHILD):
 195                 man->last->child = p;
 196                 p->parent = man->last;
 197                 break;
 198         default:
 199                 abort();
 200                 /* NOTREACHED */
 201         }
 202
 203         p->parent->nchild++;
 204
 205         if ( ! man_valid_pre(man, p))
 206                 return(0);
 207
 208         switch (p->type) {
 209         case (MAN_HEAD):
 210                 assert(MAN_BLOCK == p->parent->type);
 211                 p->parent->head = p;
 212                 break;
 213         case (MAN_BODY):
 214                 assert(MAN_BLOCK == p->parent->type);
 215                 p->parent->body = p;
 216                 break;
 217         default:
 218                 break;
 219         }
 220
 221         man->last = p;
 222
 223         switch (p->type) {
 224         case (MAN_TEXT):
 225                 if ( ! man_valid_post(man))
 226                         return(0);
 227                 if ( ! man_action_post(man))
 228                         return(0);
 229                 break;
 230         default:
 231                 break;
 232         }
 233
 234         return(1);
 235 }
 236
 237
 238 static struct man_node *
 239 man_node_alloc(int line, int pos, enum man_type type, int tok)
 240 {
 241         struct man_node *p;
 242
 243         p = mandoc_calloc(1, sizeof(struct man_node));
 244         p->line = line;
 245         p->pos = pos;
 246         p->type = type;
 247         p->tok = tok;
 248         return(p);
 249 }
 250
 251
 252 int
 253 man_elem_alloc(struct man *m, int line, int pos, int tok)
 254 {
 255         struct man_node *p;
 256
 257         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 258         if ( ! man_node_append(m, p))
 259                 return(0);
 260         m->next = MAN_NEXT_CHILD;
 261         return(1);
 262 }
 263
 264
 265 int
 266 man_head_alloc(struct man *m, int line, int pos, int tok)
 267 {
 268         struct man_node *p;
 269
 270         p = man_node_alloc(line, pos, MAN_HEAD, tok);
 271         if ( ! man_node_append(m, p))
 272                 return(0);
 273         m->next = MAN_NEXT_CHILD;
 274         return(1);
 275 }
 276
 277
 278 int
 279 man_body_alloc(struct man *m, int line, int pos, int tok)
 280 {
 281         struct man_node *p;
 282
 283         p = man_node_alloc(line, pos, MAN_BODY, tok);
 284         if ( ! man_node_append(m, p))
 285                 return(0);
 286         m->next = MAN_NEXT_CHILD;
 287         return(1);
 288 }
 289
 290
 291 int
 292 man_block_alloc(struct man *m, int line, int pos, int tok)
 293 {
 294         struct man_node *p;
 295
 296         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 297         if ( ! man_node_append(m, p))
 298                 return(0);
 299         m->next = MAN_NEXT_CHILD;
 300         return(1);
 301 }
 302
 303
 304 static int
 305 pstring(struct man *m, int line, int pos,
 306                 const char *p, size_t len)
 307 {
 308         struct man_node *n;
 309         size_t           sv;
 310
 311         n = man_node_alloc(line, pos, MAN_TEXT, -1);
 312         n->string = mandoc_malloc(len + 1);
 313         sv = strlcpy(n->string, p, len + 1);
 314
 315         /* Prohibit truncation. */
 316         assert(sv < len + 1);
 317
 318         if ( ! man_node_append(m, n))
 319                 return(0);
 320         m->next = MAN_NEXT_SIBLING;
 321         return(1);
 322 }
 323
 324
 325 int
 326 man_word_alloc(struct man *m, int line, int pos, const char *word)
 327 {
 328
 329         return(pstring(m, line, pos, word, strlen(word)));
 330 }
 331
 332
 333 void
 334 man_node_free(struct man_node *p)
 335 {
 336
 337         if (p->string)
 338                 free(p->string);
 339         if (p->parent)
 340                 p->parent->nchild--;
 341         free(p);
 342 }
 343
 344
 345 void
 346 man_node_freelist(struct man_node *p)
 347 {
 348         struct man_node *n;
 349
 350         if (p->child)
 351                 man_node_freelist(p->child);
 352         assert(0 == p->nchild);
 353         n = p->next;
 354         man_node_free(p);
 355         if (n)
 356                 man_node_freelist(n);
 357 }
 358
 359
 360 static int
 361 man_ptext(struct man *m, int line, char *buf)
 362 {
 363         int              i, j;
 364         char             sv;
 365
 366         /* Literal free-form text whitespace is preserved. */
 367
 368         if (MAN_LITERAL & m->flags) {
 369                 if ( ! man_word_alloc(m, line, 0, buf))
 370                         return(0);
 371                 goto descope;
 372         }
 373
 374         /* First de-chunk and allocate words. */
 375
 376         for (i = 0; ' ' == buf[i]; i++)
 377                 /* Skip leading whitespace. */ ;
 378
 379         if ('\0' == buf[i]) {
 380                 /* Trailing whitespace? */
 381                 if (i && ' ' == buf[i - 1])
 382                         if ( ! man_pwarn(m, line, i - 1, WTSPACE))
 383                                 return(0);
 384                 if ( ! pstring(m, line, 0, &buf[i], 0))
 385                         return(0);
 386                 goto descope;
 387         }
 388
 389         for (j = i; buf[i]; i++) {
 390                 if (' ' != buf[i])
 391                         continue;
 392
 393                 /* Escaped whitespace. */
 394                 if (i && ' ' == buf[i] && '\\' == buf[i - 1])
 395                         continue;
 396
 397                 sv = buf[i];
 398                 buf[i++] = '\0';
 399
 400                 if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 401                         return(0);
 402
 403                 /* Trailing whitespace?  Check at overwritten byte. */
 404
 405                 if (' ' == sv && '\0' == buf[i])
 406                         if ( ! man_pwarn(m, line, i - 1, WTSPACE))
 407                                 return(0);
 408
 409                 for ( ; ' ' == buf[i]; i++)
 410                         /* Skip trailing whitespace. */ ;
 411
 412                 j = i;
 413
 414                 /* Trailing whitespace? */
 415
 416                 if (' ' == buf[i - 1] && '\0' == buf[i])
 417                         if ( ! man_pwarn(m, line, i - 1, WTSPACE))
 418                                 return(0);
 419
 420                 if ('\0' == buf[i])
 421                         break;
 422         }
 423
 424         if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 425                 return(0);
 426
 427 descope:
 428
 429         /*
 430          * Co-ordinate what happens with having a next-line scope open:
 431          * first close out the element scope (if applicable), then close
 432          * out the block scope (also if applicable).
 433          */
 434
 435         if (MAN_ELINE & m->flags) {
 436                 m->flags &= ~MAN_ELINE;
 437                 if ( ! man_unscope(m, m->last->parent))
 438                         return(0);
 439         }
 440
 441         if ( ! (MAN_BLINE & m->flags))
 442                 return(1);
 443         m->flags &= ~MAN_BLINE;
 444
 445         if ( ! man_unscope(m, m->last->parent))
 446                 return(0);
 447         return(man_body_alloc(m, line, 0, m->last->tok));
 448 }
 449
 450
 451 static int
 452 macrowarn(struct man *m, int ln, const char *buf)
 453 {
 454         if ( ! (MAN_IGN_MACRO & m->pflags))
 455                 return(man_verr(m, ln, 0,
 456                                 "unknown macro: %s%s",
 457                                 buf, strlen(buf) > 3 ? "..." : ""));
 458         return(man_vwarn(m, ln, 0, "unknown macro: %s%s",
 459                                 buf, strlen(buf) > 3 ? "..." : ""));
 460 }
 461
 462
 463 int
 464 man_pmacro(struct man *m, int ln, char *buf)
 465 {
 466         int              i, j, c, ppos, fl;
 467         char             mac[5];
 468         struct man_node *n;
 469
 470         /* Comments and empties are quickly ignored. */
 471
 472         fl = m->flags;
 473
 474         if ('\0' == buf[1])
 475                 return(1);
 476
 477         i = 1;
 478
 479         if (' ' == buf[i]) {
 480                 i++;
 481                 while (buf[i] && ' ' == buf[i])
 482                         i++;
 483                 if ('\0' == buf[i])
 484                         goto out;
 485         }
 486
 487         ppos = i;
 488
 489         /* Copy the first word into a nil-terminated buffer. */
 490
 491         for (j = 0; j < 4; j++, i++) {
 492                 if ('\0' == (mac[j] = buf[i]))
 493                         break;
 494                 else if (' ' == buf[i])
 495                         break;
 496
 497                 /* Check for invalid characters. */
 498
 499                 if (isgraph((u_char)buf[i]))
 500                         continue;
 501                 return(man_perr(m, ln, i, WNPRINT));
 502         }
 503
 504         mac[j] = '\0';
 505
 506         if (j == 4 || j < 1) {
 507                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 508                         (void)man_perr(m, ln, ppos, WMACROFORM);
 509                         goto err;
 510                 }
 511                 if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
 512                         goto err;
 513                 return(1);
 514         }
 515
 516         if (MAN_MAX == (c = man_hash_find(mac))) {
 517                 if ( ! macrowarn(m, ln, mac))
 518                         goto err;
 519                 return(1);
 520         }
 521
 522         /* The macro is sane.  Jump to the next word. */
 523
 524         while (buf[i] && ' ' == buf[i])
 525                 i++;
 526
 527         /* Trailing whitespace? */
 528
 529         if ('\0' == buf[i] && ' ' == buf[i - 1])
 530                 if ( ! man_pwarn(m, ln, i - 1, WTSPACE))
 531                         goto err;
 532
 533         /* Remove prior ELINE macro, if applicable. */
 534
 535         if (m->flags & MAN_ELINE) {
 536                 n = m->last;
 537                 assert(NULL == n->child);
 538                 assert(0 == n->nchild);
 539                 if ( ! man_nwarn(m, n, WLNSCOPE))
 540                         return(0);
 541
 542                 if (n->prev) {
 543                         assert(n != n->parent->child);
 544                         assert(n == n->prev->next);
 545                         n->prev->next = NULL;
 546                         m->last = n->prev;
 547                         m->next = MAN_NEXT_SIBLING;
 548                 } else {
 549                         assert(n == n->parent->child);
 550                         n->parent->child = NULL;
 551                         m->last = n->parent;
 552                         m->next = MAN_NEXT_CHILD;
 553                 }
 554
 555                 man_node_free(n);
 556                 m->flags &= ~MAN_ELINE;
 557         }
 558
 559         /* Begin recursive parse sequence. */
 560
 561         assert(man_macros[c].fp);
 562
 563         if ( ! (*man_macros[c].fp)(m, c, ln, ppos, &i, buf))
 564                 goto err;
 565
 566 out:
 567         if ( ! (MAN_BLINE & fl))
 568                 return(1);
 569
 570         /*
 571          * If we've opened a new next-line element scope, then return
 572          * now, as the next line will close out the block scope.
 573          */
 574
 575         if (MAN_ELINE & m->flags)
 576                 return(1);
 577
 578         /* Close out the block scope opened in the prior line.  */
 579
 580         assert(MAN_BLINE & m->flags);
 581         m->flags &= ~MAN_BLINE;
 582
 583         if ( ! man_unscope(m, m->last->parent))
 584                 return(0);
 585         return(man_body_alloc(m, ln, 0, m->last->tok));
 586
 587 err:    /* Error out. */
 588
 589         m->flags |= MAN_HALT;
 590         return(0);
 591 }
 592
 593
 594 int
 595 man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
 596 {
 597         char             buf[256];
 598         va_list          ap;
 599
 600         if (NULL == man->cb.man_err)
 601                 return(0);
 602
 603         va_start(ap, fmt);
 604         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 605         va_end(ap);
 606         return((*man->cb.man_err)(man->data, ln, pos, buf));
 607 }
 608
 609
 610 int
 611 man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 612 {
 613         char             buf[256];
 614         va_list          ap;
 615
 616         if (NULL == man->cb.man_warn)
 617                 return(0);
 618
 619         va_start(ap, fmt);
 620         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 621         va_end(ap);
 622         return((*man->cb.man_warn)(man->data, ln, pos, buf));
 623 }
 624
 625
 626 int
 627 man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 628 {
 629         const char       *p;
 630
 631         p = __man_merrnames[(int)type];
 632         assert(p);
 633
 634         if (iserr)
 635                 return(man_verr(m, line, pos, p));
 636
 637         return(man_vwarn(m, line, pos, p));
 638 }