man.c

   1 /*      $Id: man.c,v 1.39 2009/08/22 09:10:38 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 <assert.h>
  18 #include <ctype.h>
  19 #include <stdarg.h>
  20 #include <stdlib.h>
  21 #include <stdio.h>
  22 #include <string.h>
  23
  24 #include "libman.h"
  25
  26 const   char *const __man_merrnames[WERRMAX] = {
  27         "invalid character", /* WNPRINT */
  28         "system: malloc error", /* WNMEM */
  29         "invalid manual section", /* WMSEC */
  30         "invalid date format", /* WDATE */
  31         "scope of prior line violated", /* WLNSCOPE */
  32         "trailing whitespace", /* WTSPACE */
  33         "unterminated quoted parameter", /* WTQUOTE */
  34         "document has no body", /* WNODATA */
  35         "document has no title/section", /* WNOTITLE */
  36         "invalid escape sequence", /* WESCAPE */
  37         "invalid number format", /* WNUMFMT */
  38         "expected block head arguments", /* WHEADARGS */
  39         "expected block body arguments", /* WBODYARGS */
  40         "expected empty block head", /* WNHEADARGS */
  41         "unknown macro", /* WMACRO */
  42         "ill-formed macro", /* WMACROFORM */
  43         "scope open on exit", /* WEXITSCOPE */
  44         "no scope context", /* WNOSCOPE */
  45         "literal context already open", /* WOLITERAL */
  46         "no literal context open" /* WNLITERAL */
  47 };
  48
  49 const   char *const __man_macronames[MAN_MAX] = {
  50         "br",           "TH",           "SH",           "SS",
  51         "TP",           "LP",           "PP",           "P",
  52         "IP",           "HP",           "SM",           "SB",
  53         "BI",           "IB",           "BR",           "RB",
  54         "R",            "B",            "I",            "IR",
  55         "RI",           "na",           "i",            "sp",
  56         "nf",           "fi",           "r",            "RE",
  57         "RS",           "DT",           "UC"
  58         };
  59
  60 const   char * const *man_macronames = __man_macronames;
  61
  62 static  struct man_node *man_node_alloc(int, int,
  63                                 enum man_type, int);
  64 static  int              man_node_append(struct man *,
  65                                 struct man_node *);
  66 static  int              man_ptext(struct man *, int, char *);
  67 static  int              man_pmacro(struct man *, int, char *);
  68 static  void             man_free1(struct man *);
  69 static  int              man_alloc1(struct man *);
  70 static  int              pstring(struct man *, int, int,
  71                                 const char *, size_t);
  72
  73 #ifdef __linux__
  74 extern  size_t            strlcpy(char *, const char *, size_t);
  75 #endif
  76
  77
  78 const struct man_node *
  79 man_node(const struct man *m)
  80 {
  81
  82         return(MAN_HALT & m->flags ? NULL : m->first);
  83 }
  84
  85
  86 const struct man_meta *
  87 man_meta(const struct man *m)
  88 {
  89
  90         return(MAN_HALT & m->flags ? NULL : &m->meta);
  91 }
  92
  93
  94 int
  95 man_reset(struct man *man)
  96 {
  97
  98         man_free1(man);
  99         return(man_alloc1(man));
 100 }
 101
 102
 103 void
 104 man_free(struct man *man)
 105 {
 106
 107         man_free1(man);
 108
 109         if (man->htab)
 110                 man_hash_free(man->htab);
 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         if (NULL == (p = calloc(1, sizeof(struct man))))
 121                 return(NULL);
 122
 123         if ( ! man_alloc1(p)) {
 124                 free(p);
 125                 return(NULL);
 126         }
 127
 128         p->data = data;
 129         p->pflags = pflags;
 130         (void)memcpy(&p->cb, cb, sizeof(struct man_cb));
 131
 132         if (NULL == (p->htab = man_hash_alloc())) {
 133                 free(p);
 134                 return(NULL);
 135         }
 136         return(p);
 137 }
 138
 139
 140 int
 141 man_endparse(struct man *m)
 142 {
 143
 144         if (MAN_HALT & m->flags)
 145                 return(0);
 146         else if (man_macroend(m))
 147                 return(1);
 148         m->flags |= MAN_HALT;
 149         return(0);
 150 }
 151
 152
 153 int
 154 man_parseln(struct man *m, int ln, char *buf)
 155 {
 156
 157         return('.' == *buf ?
 158                         man_pmacro(m, ln, buf) :
 159                         man_ptext(m, ln, buf));
 160 }
 161
 162
 163 static void
 164 man_free1(struct man *man)
 165 {
 166
 167         if (man->first)
 168                 man_node_freelist(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 }
 176
 177
 178 static int
 179 man_alloc1(struct man *m)
 180 {
 181
 182         bzero(&m->meta, sizeof(struct man_meta));
 183         m->flags = 0;
 184         m->last = calloc(1, sizeof(struct man_node));
 185         if (NULL == m->last)
 186                 return(0);
 187         m->first = m->last;
 188         m->last->type = MAN_ROOT;
 189         m->next = MAN_NEXT_CHILD;
 190         return(1);
 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         p->parent->nchild++;
 218
 219         if ( ! man_valid_pre(man, p))
 220                 return(0);
 221
 222         switch (p->type) {
 223         case (MAN_HEAD):
 224                 assert(MAN_BLOCK == p->parent->type);
 225                 p->parent->head = p;
 226                 break;
 227         case (MAN_BODY):
 228                 assert(MAN_BLOCK == p->parent->type);
 229                 p->parent->body = p;
 230                 break;
 231         default:
 232                 break;
 233         }
 234
 235         man->last = p;
 236
 237         switch (p->type) {
 238         case (MAN_TEXT):
 239                 if ( ! man_valid_post(man))
 240                         return(0);
 241                 if ( ! man_action_post(man))
 242                         return(0);
 243                 break;
 244         default:
 245                 break;
 246         }
 247
 248         return(1);
 249 }
 250
 251
 252 static struct man_node *
 253 man_node_alloc(int line, int pos, enum man_type type, int tok)
 254 {
 255         struct man_node *p;
 256
 257         p = calloc(1, sizeof(struct man_node));
 258         if (NULL == p)
 259                 return(NULL);
 260
 261         p->line = line;
 262         p->pos = pos;
 263         p->type = type;
 264         p->tok = tok;
 265         return(p);
 266 }
 267
 268
 269 int
 270 man_elem_alloc(struct man *m, int line, int pos, int tok)
 271 {
 272         struct man_node *p;
 273
 274         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 275         if (NULL == p)
 276                 return(0);
 277         if ( ! man_node_append(m, p))
 278                 return(0);
 279         m->next = MAN_NEXT_CHILD;
 280         return(1);
 281 }
 282
 283
 284 int
 285 man_head_alloc(struct man *m, int line, int pos, int tok)
 286 {
 287         struct man_node *p;
 288
 289         p = man_node_alloc(line, pos, MAN_HEAD, tok);
 290         if (NULL == p)
 291                 return(0);
 292         if ( ! man_node_append(m, p))
 293                 return(0);
 294         m->next = MAN_NEXT_CHILD;
 295         return(1);
 296 }
 297
 298
 299 int
 300 man_body_alloc(struct man *m, int line, int pos, int tok)
 301 {
 302         struct man_node *p;
 303
 304         p = man_node_alloc(line, pos, MAN_BODY, tok);
 305         if (NULL == p)
 306                 return(0);
 307         if ( ! man_node_append(m, p))
 308                 return(0);
 309         m->next = MAN_NEXT_CHILD;
 310         return(1);
 311 }
 312
 313
 314 int
 315 man_block_alloc(struct man *m, int line, int pos, int tok)
 316 {
 317         struct man_node *p;
 318
 319         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 320         if (NULL == p)
 321                 return(0);
 322         if ( ! man_node_append(m, p))
 323                 return(0);
 324         m->next = MAN_NEXT_CHILD;
 325         return(1);
 326 }
 327
 328
 329 static int
 330 pstring(struct man *m, int line, int pos,
 331                 const char *p, size_t len)
 332 {
 333         struct man_node *n;
 334         size_t           sv;
 335
 336         n = man_node_alloc(line, pos, MAN_TEXT, -1);
 337         if (NULL == n)
 338                 return(0);
 339
 340         n->string = malloc(len + 1);
 341         if (NULL == n->string) {
 342                 free(n);
 343                 return(0);
 344         }
 345
 346         sv = strlcpy(n->string, p, len + 1);
 347
 348         /* Prohibit truncation. */
 349         assert(sv < len + 1);
 350
 351         if ( ! man_node_append(m, n))
 352                 return(0);
 353         m->next = MAN_NEXT_SIBLING;
 354         return(1);
 355 }
 356
 357
 358 int
 359 man_word_alloc(struct man *m, int line, int pos, const char *word)
 360 {
 361
 362         return(pstring(m, line, pos, word, strlen(word)));
 363 }
 364
 365
 366 void
 367 man_node_free(struct man_node *p)
 368 {
 369
 370         if (p->string)
 371                 free(p->string);
 372         if (p->parent)
 373                 p->parent->nchild--;
 374         free(p);
 375 }
 376
 377
 378 void
 379 man_node_freelist(struct man_node *p)
 380 {
 381         struct man_node *n;
 382
 383         if (p->child)
 384                 man_node_freelist(p->child);
 385         assert(0 == p->nchild);
 386         n = p->next;
 387         man_node_free(p);
 388         if (n)
 389                 man_node_freelist(n);
 390 }
 391
 392
 393 static int
 394 man_ptext(struct man *m, int line, char *buf)
 395 {
 396         int              i, j;
 397
 398         /* Literal free-form text whitespace is preserved. */
 399
 400         if (MAN_LITERAL & m->flags) {
 401                 if ( ! man_word_alloc(m, line, 0, buf))
 402                         return(0);
 403                 goto descope;
 404         }
 405
 406         /* First de-chunk and allocate words. */
 407
 408         for (i = 0; ' ' == buf[i]; i++)
 409                 /* Skip leading whitespace. */ ;
 410         if (0 == buf[i]) {
 411                 if ( ! pstring(m, line, 0, &buf[i], 0))
 412                         return(0);
 413                 goto descope;
 414         }
 415
 416         for (j = i; buf[i]; i++) {
 417                 if (' ' != buf[i])
 418                         continue;
 419
 420                 /* Escaped whitespace. */
 421                 if (i && ' ' == buf[i] && '\\' == buf[i - 1])
 422                         continue;
 423
 424                 buf[i++] = 0;
 425                 if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 426                         return(0);
 427
 428                 for ( ; ' ' == buf[i]; i++)
 429                         /* Skip trailing whitespace. */ ;
 430
 431                 j = i;
 432                 if (0 == buf[i])
 433                         break;
 434         }
 435
 436         if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 437                 return(0);
 438
 439 descope:
 440
 441         /*
 442          * Co-ordinate what happens with having a next-line scope open:
 443          * first close out the element scope (if applicable), then close
 444          * out the block scope (also if applicable).
 445          */
 446
 447         if (MAN_ELINE & m->flags) {
 448                 m->flags &= ~MAN_ELINE;
 449                 if ( ! man_unscope(m, m->last->parent))
 450                         return(0);
 451         }
 452
 453         if ( ! (MAN_BLINE & m->flags))
 454                 return(1);
 455         m->flags &= ~MAN_BLINE;
 456
 457         if ( ! man_unscope(m, m->last->parent))
 458                 return(0);
 459         return(man_body_alloc(m, line, 0, m->last->tok));
 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                 goto out;
 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(m->htab, mac))) {
 517                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 518                         (void)man_perr(m, ln, ppos, WMACRO);
 519                         goto err;
 520                 }
 521                 if ( ! man_pwarn(m, ln, ppos, WMACRO))
 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 }