usr.bin/mandoc/man.c

   1 /*      $Id: man.c,v 1.15 2009/10/27 21:40:07 schwarze 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
  28 const   char *const __man_merrnames[WERRMAX] = {
  29         "invalid character", /* WNPRINT */
  30         "system: malloc error", /* WNMEM */
  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  int              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 int
  93 man_reset(struct man *man)
  94 {
  95
  96         man_free1(man);
  97         return(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         if (NULL == (p = calloc(1, sizeof(struct man))))
 116                 return(NULL);
 117
 118         if ( ! man_alloc1(p)) {
 119                 free(p);
 120                 return(NULL);
 121         }
 122
 123         man_hash_init();
 124
 125         p->data = data;
 126         p->pflags = pflags;
 127         (void)memcpy(&p->cb, cb, sizeof(struct man_cb));
 128         return(p);
 129 }
 130
 131
 132 int
 133 man_endparse(struct man *m)
 134 {
 135
 136         if (MAN_HALT & m->flags)
 137                 return(0);
 138         else if (man_macroend(m))
 139                 return(1);
 140         m->flags |= MAN_HALT;
 141         return(0);
 142 }
 143
 144
 145 int
 146 man_parseln(struct man *m, int ln, char *buf)
 147 {
 148
 149         return('.' == *buf ?
 150                         man_pmacro(m, ln, buf) :
 151                         man_ptext(m, ln, buf));
 152 }
 153
 154
 155 static void
 156 man_free1(struct man *man)
 157 {
 158
 159         if (man->first)
 160                 man_node_freelist(man->first);
 161         if (man->meta.title)
 162                 free(man->meta.title);
 163         if (man->meta.source)
 164                 free(man->meta.source);
 165         if (man->meta.vol)
 166                 free(man->meta.vol);
 167 }
 168
 169
 170 static int
 171 man_alloc1(struct man *m)
 172 {
 173
 174         bzero(&m->meta, sizeof(struct man_meta));
 175         m->flags = 0;
 176         m->last = calloc(1, sizeof(struct man_node));
 177         if (NULL == m->last)
 178                 return(0);
 179         m->first = m->last;
 180         m->last->type = MAN_ROOT;
 181         m->next = MAN_NEXT_CHILD;
 182         return(1);
 183 }
 184
 185
 186 static int
 187 man_node_append(struct man *man, struct man_node *p)
 188 {
 189
 190         assert(man->last);
 191         assert(man->first);
 192         assert(MAN_ROOT != p->type);
 193
 194         switch (man->next) {
 195         case (MAN_NEXT_SIBLING):
 196                 man->last->next = p;
 197                 p->prev = man->last;
 198                 p->parent = man->last->parent;
 199                 break;
 200         case (MAN_NEXT_CHILD):
 201                 man->last->child = p;
 202                 p->parent = man->last;
 203                 break;
 204         default:
 205                 abort();
 206                 /* NOTREACHED */
 207         }
 208
 209         p->parent->nchild++;
 210
 211         if ( ! man_valid_pre(man, p))
 212                 return(0);
 213
 214         switch (p->type) {
 215         case (MAN_HEAD):
 216                 assert(MAN_BLOCK == p->parent->type);
 217                 p->parent->head = p;
 218                 break;
 219         case (MAN_BODY):
 220                 assert(MAN_BLOCK == p->parent->type);
 221                 p->parent->body = p;
 222                 break;
 223         default:
 224                 break;
 225         }
 226
 227         man->last = p;
 228
 229         switch (p->type) {
 230         case (MAN_TEXT):
 231                 if ( ! man_valid_post(man))
 232                         return(0);
 233                 if ( ! man_action_post(man))
 234                         return(0);
 235                 break;
 236         default:
 237                 break;
 238         }
 239
 240         return(1);
 241 }
 242
 243
 244 static struct man_node *
 245 man_node_alloc(int line, int pos, enum man_type type, int tok)
 246 {
 247         struct man_node *p;
 248
 249         p = calloc(1, sizeof(struct man_node));
 250         if (NULL == p)
 251                 return(NULL);
 252
 253         p->line = line;
 254         p->pos = pos;
 255         p->type = type;
 256         p->tok = tok;
 257         return(p);
 258 }
 259
 260
 261 int
 262 man_elem_alloc(struct man *m, int line, int pos, int tok)
 263 {
 264         struct man_node *p;
 265
 266         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 267         if (NULL == p)
 268                 return(0);
 269         if ( ! man_node_append(m, p))
 270                 return(0);
 271         m->next = MAN_NEXT_CHILD;
 272         return(1);
 273 }
 274
 275
 276 int
 277 man_head_alloc(struct man *m, int line, int pos, int tok)
 278 {
 279         struct man_node *p;
 280
 281         p = man_node_alloc(line, pos, MAN_HEAD, tok);
 282         if (NULL == p)
 283                 return(0);
 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_body_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_BODY, tok);
 297         if (NULL == p)
 298                 return(0);
 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, int tok)
 308 {
 309         struct man_node *p;
 310
 311         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 312         if (NULL == p)
 313                 return(0);
 314         if ( ! man_node_append(m, p))
 315                 return(0);
 316         m->next = MAN_NEXT_CHILD;
 317         return(1);
 318 }
 319
 320
 321 static int
 322 pstring(struct man *m, int line, int pos,
 323                 const char *p, size_t len)
 324 {
 325         struct man_node *n;
 326         size_t           sv;
 327
 328         n = man_node_alloc(line, pos, MAN_TEXT, -1);
 329         if (NULL == n)
 330                 return(0);
 331
 332         n->string = malloc(len + 1);
 333         if (NULL == n->string) {
 334                 free(n);
 335                 return(0);
 336         }
 337
 338         sv = strlcpy(n->string, p, len + 1);
 339
 340         /* Prohibit truncation. */
 341         assert(sv < len + 1);
 342
 343         if ( ! man_node_append(m, n))
 344                 return(0);
 345         m->next = MAN_NEXT_SIBLING;
 346         return(1);
 347 }
 348
 349
 350 int
 351 man_word_alloc(struct man *m, int line, int pos, const char *word)
 352 {
 353
 354         return(pstring(m, line, pos, word, strlen(word)));
 355 }
 356
 357
 358 void
 359 man_node_free(struct man_node *p)
 360 {
 361
 362         if (p->string)
 363                 free(p->string);
 364         if (p->parent)
 365                 p->parent->nchild--;
 366         free(p);
 367 }
 368
 369
 370 void
 371 man_node_freelist(struct man_node *p)
 372 {
 373         struct man_node *n;
 374
 375         if (p->child)
 376                 man_node_freelist(p->child);
 377         assert(0 == p->nchild);
 378         n = p->next;
 379         man_node_free(p);
 380         if (n)
 381                 man_node_freelist(n);
 382 }
 383
 384
 385 static int
 386 man_ptext(struct man *m, int line, char *buf)
 387 {
 388         int              i, j;
 389
 390         /* Literal free-form text whitespace is preserved. */
 391
 392         if (MAN_LITERAL & m->flags) {
 393                 if ( ! man_word_alloc(m, line, 0, buf))
 394                         return(0);
 395                 goto descope;
 396         }
 397
 398         /* First de-chunk and allocate words. */
 399
 400         for (i = 0; ' ' == buf[i]; i++)
 401                 /* Skip leading whitespace. */ ;
 402         if (0 == buf[i]) {
 403                 if ( ! pstring(m, line, 0, &buf[i], 0))
 404                         return(0);
 405                 goto descope;
 406         }
 407
 408         for (j = i; buf[i]; i++) {
 409                 if (' ' != buf[i])
 410                         continue;
 411
 412                 /* Escaped whitespace. */
 413                 if (i && ' ' == buf[i] && '\\' == buf[i - 1])
 414                         continue;
 415
 416                 buf[i++] = 0;
 417                 if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 418                         return(0);
 419
 420                 for ( ; ' ' == buf[i]; i++)
 421                         /* Skip trailing whitespace. */ ;
 422
 423                 j = i;
 424                 if (0 == buf[i])
 425                         break;
 426         }
 427
 428         if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 429                 return(0);
 430
 431 descope:
 432
 433         /*
 434          * Co-ordinate what happens with having a next-line scope open:
 435          * first close out the element scope (if applicable), then close
 436          * out the block scope (also if applicable).
 437          */
 438
 439         if (MAN_ELINE & m->flags) {
 440                 m->flags &= ~MAN_ELINE;
 441                 if ( ! man_unscope(m, m->last->parent))
 442                         return(0);
 443         }
 444
 445         if ( ! (MAN_BLINE & m->flags))
 446                 return(1);
 447         m->flags &= ~MAN_BLINE;
 448
 449         if ( ! man_unscope(m, m->last->parent))
 450                 return(0);
 451         return(man_body_alloc(m, line, 0, m->last->tok));
 452 }
 453
 454
 455 static int
 456 macrowarn(struct man *m, int ln, const char *buf)
 457 {
 458         if ( ! (MAN_IGN_MACRO & m->pflags))
 459                 return(man_verr(m, ln, 0,
 460                                 "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, c, ppos, fl;
 471         char             mac[5];
 472         struct man_node *n;
 473
 474         /* Comments and empties are quickly ignored. */
 475
 476         fl = m->flags;
 477
 478         if (0 == buf[1])
 479                 goto out;
 480
 481         i = 1;
 482
 483         if (' ' == buf[i]) {
 484                 i++;
 485                 while (buf[i] && ' ' == buf[i])
 486                         i++;
 487                 if (0 == buf[i])
 488                         goto out;
 489         }
 490
 491         ppos = i;
 492
 493         /* Copy the first word into a nil-terminated buffer. */
 494
 495         for (j = 0; j < 4; j++, i++) {
 496                 if (0 == (mac[j] = buf[i]))
 497                         break;
 498                 else if (' ' == buf[i])
 499                         break;
 500
 501                 /* Check for invalid characters. */
 502
 503                 if (isgraph((u_char)buf[i]))
 504                         continue;
 505                 return(man_perr(m, ln, i, WNPRINT));
 506         }
 507
 508         mac[j] = 0;
 509
 510         if (j == 4 || j < 1) {
 511                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 512                         (void)man_perr(m, ln, ppos, WMACROFORM);
 513                         goto err;
 514                 }
 515                 if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
 516                         goto err;
 517                 return(1);
 518         }
 519
 520         if (MAN_MAX == (c = man_hash_find(mac))) {
 521                 if ( ! macrowarn(m, ln, mac))
 522                         goto err;
 523                 return(1);
 524         }
 525
 526         /* The macro is sane.  Jump to the next word. */
 527
 528         while (buf[i] && ' ' == buf[i])
 529                 i++;
 530
 531         /* Remove prior ELINE macro, if applicable. */
 532
 533         if (m->flags & MAN_ELINE) {
 534                 n = m->last;
 535                 assert(NULL == n->child);
 536                 assert(0 == n->nchild);
 537                 if ( ! man_nwarn(m, n, WLNSCOPE))
 538                         return(0);
 539
 540                 if (n->prev) {
 541                         assert(n != n->parent->child);
 542                         assert(n == n->prev->next);
 543                         n->prev->next = NULL;
 544                         m->last = n->prev;
 545                         m->next = MAN_NEXT_SIBLING;
 546                 } else {
 547                         assert(n == n->parent->child);
 548                         n->parent->child = NULL;
 549                         m->last = n->parent;
 550                         m->next = MAN_NEXT_CHILD;
 551                 }
 552
 553                 man_node_free(n);
 554                 m->flags &= ~MAN_ELINE;
 555         }
 556
 557         /* Begin recursive parse sequence. */
 558
 559         assert(man_macros[c].fp);
 560
 561         if ( ! (*man_macros[c].fp)(m, c, ln, ppos, &i, buf))
 562                 goto err;
 563
 564 out:
 565         if ( ! (MAN_BLINE & fl))
 566                 return(1);
 567
 568         /*
 569          * If we've opened a new next-line element scope, then return
 570          * now, as the next line will close out the block scope.
 571          */
 572
 573         if (MAN_ELINE & m->flags)
 574                 return(1);
 575
 576         /* Close out the block scope opened in the prior line.  */
 577
 578         assert(MAN_BLINE & m->flags);
 579         m->flags &= ~MAN_BLINE;
 580
 581         if ( ! man_unscope(m, m->last->parent))
 582                 return(0);
 583         return(man_body_alloc(m, ln, 0, m->last->tok));
 584
 585 err:    /* Error out. */
 586
 587         m->flags |= MAN_HALT;
 588         return(0);
 589 }
 590
 591
 592 int
 593 man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
 594 {
 595         char             buf[256];
 596         va_list          ap;
 597
 598         if (NULL == man->cb.man_err)
 599                 return(0);
 600
 601         va_start(ap, fmt);
 602         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 603         va_end(ap);
 604         return((*man->cb.man_err)(man->data, ln, pos, buf));
 605 }
 606
 607
 608 int
 609 man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 610 {
 611         char             buf[256];
 612         va_list          ap;
 613
 614         if (NULL == man->cb.man_warn)
 615                 return(0);
 616
 617         va_start(ap, fmt);
 618         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 619         va_end(ap);
 620         return((*man->cb.man_warn)(man->data, ln, pos, buf));
 621 }
 622
 623
 624 int
 625 man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 626 {
 627         const char       *p;
 628
 629         p = __man_merrnames[(int)type];
 630         assert(p);
 631
 632         if (iserr)
 633                 return(man_verr(m, line, pos, p));
 634
 635         return(man_vwarn(m, line, pos, p));
 636 }