man.c

   1 /*      $Id: man.c,v 1.40 2009/09/16 14:40:56 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         free(man);
 109 }
 110
 111
 112 struct man *
 113 man_alloc(void *data, int pflags, const struct man_cb *cb)
 114 {
 115         struct man      *p;
 116
 117         if (NULL == (p = calloc(1, sizeof(struct man))))
 118                 return(NULL);
 119
 120         if ( ! man_alloc1(p)) {
 121                 free(p);
 122                 return(NULL);
 123         }
 124
 125         man_hash_init();
 126
 127         p->data = data;
 128         p->pflags = pflags;
 129         (void)memcpy(&p->cb, cb, sizeof(struct man_cb));
 130         return(p);
 131 }
 132
 133
 134 int
 135 man_endparse(struct man *m)
 136 {
 137
 138         if (MAN_HALT & m->flags)
 139                 return(0);
 140         else if (man_macroend(m))
 141                 return(1);
 142         m->flags |= MAN_HALT;
 143         return(0);
 144 }
 145
 146
 147 int
 148 man_parseln(struct man *m, int ln, char *buf)
 149 {
 150
 151         return('.' == *buf ?
 152                         man_pmacro(m, ln, buf) :
 153                         man_ptext(m, ln, buf));
 154 }
 155
 156
 157 static void
 158 man_free1(struct man *man)
 159 {
 160
 161         if (man->first)
 162                 man_node_freelist(man->first);
 163         if (man->meta.title)
 164                 free(man->meta.title);
 165         if (man->meta.source)
 166                 free(man->meta.source);
 167         if (man->meta.vol)
 168                 free(man->meta.vol);
 169 }
 170
 171
 172 static int
 173 man_alloc1(struct man *m)
 174 {
 175
 176         bzero(&m->meta, sizeof(struct man_meta));
 177         m->flags = 0;
 178         m->last = calloc(1, sizeof(struct man_node));
 179         if (NULL == m->last)
 180                 return(0);
 181         m->first = m->last;
 182         m->last->type = MAN_ROOT;
 183         m->next = MAN_NEXT_CHILD;
 184         return(1);
 185 }
 186
 187
 188 static int
 189 man_node_append(struct man *man, struct man_node *p)
 190 {
 191
 192         assert(man->last);
 193         assert(man->first);
 194         assert(MAN_ROOT != p->type);
 195
 196         switch (man->next) {
 197         case (MAN_NEXT_SIBLING):
 198                 man->last->next = p;
 199                 p->prev = man->last;
 200                 p->parent = man->last->parent;
 201                 break;
 202         case (MAN_NEXT_CHILD):
 203                 man->last->child = p;
 204                 p->parent = man->last;
 205                 break;
 206         default:
 207                 abort();
 208                 /* NOTREACHED */
 209         }
 210
 211         p->parent->nchild++;
 212
 213         if ( ! man_valid_pre(man, p))
 214                 return(0);
 215
 216         switch (p->type) {
 217         case (MAN_HEAD):
 218                 assert(MAN_BLOCK == p->parent->type);
 219                 p->parent->head = p;
 220                 break;
 221         case (MAN_BODY):
 222                 assert(MAN_BLOCK == p->parent->type);
 223                 p->parent->body = p;
 224                 break;
 225         default:
 226                 break;
 227         }
 228
 229         man->last = p;
 230
 231         switch (p->type) {
 232         case (MAN_TEXT):
 233                 if ( ! man_valid_post(man))
 234                         return(0);
 235                 if ( ! man_action_post(man))
 236                         return(0);
 237                 break;
 238         default:
 239                 break;
 240         }
 241
 242         return(1);
 243 }
 244
 245
 246 static struct man_node *
 247 man_node_alloc(int line, int pos, enum man_type type, int tok)
 248 {
 249         struct man_node *p;
 250
 251         p = calloc(1, sizeof(struct man_node));
 252         if (NULL == p)
 253                 return(NULL);
 254
 255         p->line = line;
 256         p->pos = pos;
 257         p->type = type;
 258         p->tok = tok;
 259         return(p);
 260 }
 261
 262
 263 int
 264 man_elem_alloc(struct man *m, int line, int pos, int tok)
 265 {
 266         struct man_node *p;
 267
 268         p = man_node_alloc(line, pos, MAN_ELEM, tok);
 269         if (NULL == p)
 270                 return(0);
 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_head_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_HEAD, tok);
 284         if (NULL == p)
 285                 return(0);
 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, int tok)
 295 {
 296         struct man_node *p;
 297
 298         p = man_node_alloc(line, pos, MAN_BODY, tok);
 299         if (NULL == p)
 300                 return(0);
 301         if ( ! man_node_append(m, p))
 302                 return(0);
 303         m->next = MAN_NEXT_CHILD;
 304         return(1);
 305 }
 306
 307
 308 int
 309 man_block_alloc(struct man *m, int line, int pos, int tok)
 310 {
 311         struct man_node *p;
 312
 313         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 314         if (NULL == p)
 315                 return(0);
 316         if ( ! man_node_append(m, p))
 317                 return(0);
 318         m->next = MAN_NEXT_CHILD;
 319         return(1);
 320 }
 321
 322
 323 static int
 324 pstring(struct man *m, int line, int pos,
 325                 const char *p, size_t len)
 326 {
 327         struct man_node *n;
 328         size_t           sv;
 329
 330         n = man_node_alloc(line, pos, MAN_TEXT, -1);
 331         if (NULL == n)
 332                 return(0);
 333
 334         n->string = malloc(len + 1);
 335         if (NULL == n->string) {
 336                 free(n);
 337                 return(0);
 338         }
 339
 340         sv = strlcpy(n->string, p, len + 1);
 341
 342         /* Prohibit truncation. */
 343         assert(sv < len + 1);
 344
 345         if ( ! man_node_append(m, n))
 346                 return(0);
 347         m->next = MAN_NEXT_SIBLING;
 348         return(1);
 349 }
 350
 351
 352 int
 353 man_word_alloc(struct man *m, int line, int pos, const char *word)
 354 {
 355
 356         return(pstring(m, line, pos, word, strlen(word)));
 357 }
 358
 359
 360 void
 361 man_node_free(struct man_node *p)
 362 {
 363
 364         if (p->string)
 365                 free(p->string);
 366         if (p->parent)
 367                 p->parent->nchild--;
 368         free(p);
 369 }
 370
 371
 372 void
 373 man_node_freelist(struct man_node *p)
 374 {
 375         struct man_node *n;
 376
 377         if (p->child)
 378                 man_node_freelist(p->child);
 379         assert(0 == p->nchild);
 380         n = p->next;
 381         man_node_free(p);
 382         if (n)
 383                 man_node_freelist(n);
 384 }
 385
 386
 387 static int
 388 man_ptext(struct man *m, int line, char *buf)
 389 {
 390         int              i, j;
 391
 392         /* Literal free-form text whitespace is preserved. */
 393
 394         if (MAN_LITERAL & m->flags) {
 395                 if ( ! man_word_alloc(m, line, 0, buf))
 396                         return(0);
 397                 goto descope;
 398         }
 399
 400         /* First de-chunk and allocate words. */
 401
 402         for (i = 0; ' ' == buf[i]; i++)
 403                 /* Skip leading whitespace. */ ;
 404         if (0 == buf[i]) {
 405                 if ( ! pstring(m, line, 0, &buf[i], 0))
 406                         return(0);
 407                 goto descope;
 408         }
 409
 410         for (j = i; buf[i]; i++) {
 411                 if (' ' != buf[i])
 412                         continue;
 413
 414                 /* Escaped whitespace. */
 415                 if (i && ' ' == buf[i] && '\\' == buf[i - 1])
 416                         continue;
 417
 418                 buf[i++] = 0;
 419                 if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 420                         return(0);
 421
 422                 for ( ; ' ' == buf[i]; i++)
 423                         /* Skip trailing whitespace. */ ;
 424
 425                 j = i;
 426                 if (0 == buf[i])
 427                         break;
 428         }
 429
 430         if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 431                 return(0);
 432
 433 descope:
 434
 435         /*
 436          * Co-ordinate what happens with having a next-line scope open:
 437          * first close out the element scope (if applicable), then close
 438          * out the block scope (also if applicable).
 439          */
 440
 441         if (MAN_ELINE & m->flags) {
 442                 m->flags &= ~MAN_ELINE;
 443                 if ( ! man_unscope(m, m->last->parent))
 444                         return(0);
 445         }
 446
 447         if ( ! (MAN_BLINE & m->flags))
 448                 return(1);
 449         m->flags &= ~MAN_BLINE;
 450
 451         if ( ! man_unscope(m, m->last->parent))
 452                 return(0);
 453         return(man_body_alloc(m, line, 0, m->last->tok));
 454 }
 455
 456
 457 int
 458 man_pmacro(struct man *m, int ln, char *buf)
 459 {
 460         int              i, j, c, ppos, fl;
 461         char             mac[5];
 462         struct man_node *n;
 463
 464         /* Comments and empties are quickly ignored. */
 465
 466         fl = m->flags;
 467
 468         if (0 == buf[1])
 469                 goto out;
 470
 471         i = 1;
 472
 473         if (' ' == buf[i]) {
 474                 i++;
 475                 while (buf[i] && ' ' == buf[i])
 476                         i++;
 477                 if (0 == buf[i])
 478                         goto out;
 479         }
 480
 481         ppos = i;
 482
 483         /* Copy the first word into a nil-terminated buffer. */
 484
 485         for (j = 0; j < 4; j++, i++) {
 486                 if (0 == (mac[j] = buf[i]))
 487                         break;
 488                 else if (' ' == buf[i])
 489                         break;
 490
 491                 /* Check for invalid characters. */
 492
 493                 if (isgraph((u_char)buf[i]))
 494                         continue;
 495                 return(man_perr(m, ln, i, WNPRINT));
 496         }
 497
 498         mac[j] = 0;
 499
 500         if (j == 4 || j < 1) {
 501                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 502                         (void)man_perr(m, ln, ppos, WMACROFORM);
 503                         goto err;
 504                 }
 505                 if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
 506                         goto err;
 507                 return(1);
 508         }
 509
 510         if (MAN_MAX == (c = man_hash_find(mac))) {
 511                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 512                         (void)man_perr(m, ln, ppos, WMACRO);
 513                         goto err;
 514                 }
 515                 if ( ! man_pwarn(m, ln, ppos, WMACRO))
 516                         goto err;
 517                 return(1);
 518         }
 519
 520         /* The macro is sane.  Jump to the next word. */
 521
 522         while (buf[i] && ' ' == buf[i])
 523                 i++;
 524
 525         /* Remove prior ELINE macro, if applicable. */
 526
 527         if (m->flags & MAN_ELINE) {
 528                 n = m->last;
 529                 assert(NULL == n->child);
 530                 assert(0 == n->nchild);
 531                 if ( ! man_nwarn(m, n, WLNSCOPE))
 532                         return(0);
 533
 534                 if (n->prev) {
 535                         assert(n != n->parent->child);
 536                         assert(n == n->prev->next);
 537                         n->prev->next = NULL;
 538                         m->last = n->prev;
 539                         m->next = MAN_NEXT_SIBLING;
 540                 } else {
 541                         assert(n == n->parent->child);
 542                         n->parent->child = NULL;
 543                         m->last = n->parent;
 544                         m->next = MAN_NEXT_CHILD;
 545                 }
 546
 547                 man_node_free(n);
 548                 m->flags &= ~MAN_ELINE;
 549         }
 550
 551         /* Begin recursive parse sequence. */
 552
 553         assert(man_macros[c].fp);
 554
 555         if ( ! (*man_macros[c].fp)(m, c, ln, ppos, &i, buf))
 556                 goto err;
 557
 558 out:
 559         if ( ! (MAN_BLINE & fl))
 560                 return(1);
 561
 562         /*
 563          * If we've opened a new next-line element scope, then return
 564          * now, as the next line will close out the block scope.
 565          */
 566
 567         if (MAN_ELINE & m->flags)
 568                 return(1);
 569
 570         /* Close out the block scope opened in the prior line.  */
 571
 572         assert(MAN_BLINE & m->flags);
 573         m->flags &= ~MAN_BLINE;
 574
 575         if ( ! man_unscope(m, m->last->parent))
 576                 return(0);
 577         return(man_body_alloc(m, ln, 0, m->last->tok));
 578
 579 err:    /* Error out. */
 580
 581         m->flags |= MAN_HALT;
 582         return(0);
 583 }
 584
 585
 586 int
 587 man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
 588 {
 589         char             buf[256];
 590         va_list          ap;
 591
 592         if (NULL == man->cb.man_err)
 593                 return(0);
 594
 595         va_start(ap, fmt);
 596         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 597         va_end(ap);
 598         return((*man->cb.man_err)(man->data, ln, pos, buf));
 599 }
 600
 601
 602 int
 603 man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 604 {
 605         char             buf[256];
 606         va_list          ap;
 607
 608         if (NULL == man->cb.man_warn)
 609                 return(0);
 610
 611         va_start(ap, fmt);
 612         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 613         va_end(ap);
 614         return((*man->cb.man_warn)(man->data, ln, pos, buf));
 615 }
 616
 617
 618 int
 619 man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 620 {
 621         const char       *p;
 622
 623         p = __man_merrnames[(int)type];
 624         assert(p);
 625
 626         if (iserr)
 627                 return(man_verr(m, line, pos, p));
 628
 629         return(man_vwarn(m, line, pos, p));
 630 }