man.c

   1 /*      $Id: man.c,v 1.44 2009/10/27 08:26:12 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
  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 #ifdef __linux__
  76 extern  size_t            strlcpy(char *, const char *, size_t);
  77 #endif
  78
  79
  80 const struct man_node *
  81 man_node(const struct man *m)
  82 {
  83
  84         return(MAN_HALT & m->flags ? NULL : m->first);
  85 }
  86
  87
  88 const struct man_meta *
  89 man_meta(const struct man *m)
  90 {
  91
  92         return(MAN_HALT & m->flags ? NULL : &m->meta);
  93 }
  94
  95
  96 int
  97 man_reset(struct man *man)
  98 {
  99
 100         man_free1(man);
 101         return(man_alloc1(man));
 102 }
 103
 104
 105 void
 106 man_free(struct man *man)
 107 {
 108
 109         man_free1(man);
 110         free(man);
 111 }
 112
 113
 114 struct man *
 115 man_alloc(void *data, int pflags, const struct man_cb *cb)
 116 {
 117         struct man      *p;
 118
 119         if (NULL == (p = calloc(1, sizeof(struct man))))
 120                 return(NULL);
 121
 122         if ( ! man_alloc1(p)) {
 123                 free(p);
 124                 return(NULL);
 125         }
 126
 127         man_hash_init();
 128
 129         p->data = data;
 130         p->pflags = pflags;
 131         (void)memcpy(&p->cb, cb, sizeof(struct man_cb));
 132         return(p);
 133 }
 134
 135
 136 int
 137 man_endparse(struct man *m)
 138 {
 139
 140         if (MAN_HALT & m->flags)
 141                 return(0);
 142         else if (man_macroend(m))
 143                 return(1);
 144         m->flags |= MAN_HALT;
 145         return(0);
 146 }
 147
 148
 149 int
 150 man_parseln(struct man *m, int ln, char *buf)
 151 {
 152
 153         return('.' == *buf ?
 154                         man_pmacro(m, ln, buf) :
 155                         man_ptext(m, ln, buf));
 156 }
 157
 158
 159 static void
 160 man_free1(struct man *man)
 161 {
 162
 163         if (man->first)
 164                 man_node_freelist(man->first);
 165         if (man->meta.title)
 166                 free(man->meta.title);
 167         if (man->meta.source)
 168                 free(man->meta.source);
 169         if (man->meta.vol)
 170                 free(man->meta.vol);
 171 }
 172
 173
 174 static int
 175 man_alloc1(struct man *m)
 176 {
 177
 178         memset(&m->meta, 0, sizeof(struct man_meta));
 179         m->flags = 0;
 180         m->last = calloc(1, sizeof(struct man_node));
 181         if (NULL == m->last)
 182                 return(0);
 183         m->first = m->last;
 184         m->last->type = MAN_ROOT;
 185         m->next = MAN_NEXT_CHILD;
 186         return(1);
 187 }
 188
 189
 190 static int
 191 man_node_append(struct man *man, struct man_node *p)
 192 {
 193
 194         assert(man->last);
 195         assert(man->first);
 196         assert(MAN_ROOT != p->type);
 197
 198         switch (man->next) {
 199         case (MAN_NEXT_SIBLING):
 200                 man->last->next = p;
 201                 p->prev = man->last;
 202                 p->parent = man->last->parent;
 203                 break;
 204         case (MAN_NEXT_CHILD):
 205                 man->last->child = p;
 206                 p->parent = man->last;
 207                 break;
 208         default:
 209                 abort();
 210                 /* NOTREACHED */
 211         }
 212
 213         p->parent->nchild++;
 214
 215         if ( ! man_valid_pre(man, p))
 216                 return(0);
 217
 218         switch (p->type) {
 219         case (MAN_HEAD):
 220                 assert(MAN_BLOCK == p->parent->type);
 221                 p->parent->head = p;
 222                 break;
 223         case (MAN_BODY):
 224                 assert(MAN_BLOCK == p->parent->type);
 225                 p->parent->body = p;
 226                 break;
 227         default:
 228                 break;
 229         }
 230
 231         man->last = p;
 232
 233         switch (p->type) {
 234         case (MAN_TEXT):
 235                 if ( ! man_valid_post(man))
 236                         return(0);
 237                 if ( ! man_action_post(man))
 238                         return(0);
 239                 break;
 240         default:
 241                 break;
 242         }
 243
 244         return(1);
 245 }
 246
 247
 248 static struct man_node *
 249 man_node_alloc(int line, int pos, enum man_type type, int tok)
 250 {
 251         struct man_node *p;
 252
 253         p = calloc(1, sizeof(struct man_node));
 254         if (NULL == p)
 255                 return(NULL);
 256
 257         p->line = line;
 258         p->pos = pos;
 259         p->type = type;
 260         p->tok = tok;
 261         return(p);
 262 }
 263
 264
 265 int
 266 man_elem_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_ELEM, tok);
 271         if (NULL == p)
 272                 return(0);
 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, int tok)
 282 {
 283         struct man_node *p;
 284
 285         p = man_node_alloc(line, pos, MAN_HEAD, tok);
 286         if (NULL == p)
 287                 return(0);
 288         if ( ! man_node_append(m, p))
 289                 return(0);
 290         m->next = MAN_NEXT_CHILD;
 291         return(1);
 292 }
 293
 294
 295 int
 296 man_body_alloc(struct man *m, int line, int pos, int tok)
 297 {
 298         struct man_node *p;
 299
 300         p = man_node_alloc(line, pos, MAN_BODY, tok);
 301         if (NULL == p)
 302                 return(0);
 303         if ( ! man_node_append(m, p))
 304                 return(0);
 305         m->next = MAN_NEXT_CHILD;
 306         return(1);
 307 }
 308
 309
 310 int
 311 man_block_alloc(struct man *m, int line, int pos, int tok)
 312 {
 313         struct man_node *p;
 314
 315         p = man_node_alloc(line, pos, MAN_BLOCK, tok);
 316         if (NULL == p)
 317                 return(0);
 318         if ( ! man_node_append(m, p))
 319                 return(0);
 320         m->next = MAN_NEXT_CHILD;
 321         return(1);
 322 }
 323
 324
 325 static int
 326 pstring(struct man *m, int line, int pos,
 327                 const char *p, size_t len)
 328 {
 329         struct man_node *n;
 330         size_t           sv;
 331
 332         n = man_node_alloc(line, pos, MAN_TEXT, -1);
 333         if (NULL == n)
 334                 return(0);
 335
 336         n->string = malloc(len + 1);
 337         if (NULL == n->string) {
 338                 free(n);
 339                 return(0);
 340         }
 341
 342         sv = strlcpy(n->string, p, len + 1);
 343
 344         /* Prohibit truncation. */
 345         assert(sv < len + 1);
 346
 347         if ( ! man_node_append(m, n))
 348                 return(0);
 349         m->next = MAN_NEXT_SIBLING;
 350         return(1);
 351 }
 352
 353
 354 int
 355 man_word_alloc(struct man *m, int line, int pos, const char *word)
 356 {
 357
 358         return(pstring(m, line, pos, word, strlen(word)));
 359 }
 360
 361
 362 void
 363 man_node_free(struct man_node *p)
 364 {
 365
 366         if (p->string)
 367                 free(p->string);
 368         if (p->parent)
 369                 p->parent->nchild--;
 370         free(p);
 371 }
 372
 373
 374 void
 375 man_node_freelist(struct man_node *p)
 376 {
 377         struct man_node *n;
 378
 379         if (p->child)
 380                 man_node_freelist(p->child);
 381         assert(0 == p->nchild);
 382         n = p->next;
 383         man_node_free(p);
 384         if (n)
 385                 man_node_freelist(n);
 386 }
 387
 388
 389 static int
 390 man_ptext(struct man *m, int line, char *buf)
 391 {
 392         int              i, j;
 393
 394         /* Literal free-form text whitespace is preserved. */
 395
 396         if (MAN_LITERAL & m->flags) {
 397                 if ( ! man_word_alloc(m, line, 0, buf))
 398                         return(0);
 399                 goto descope;
 400         }
 401
 402         /* First de-chunk and allocate words. */
 403
 404         for (i = 0; ' ' == buf[i]; i++)
 405                 /* Skip leading whitespace. */ ;
 406         if (0 == buf[i]) {
 407                 if ( ! pstring(m, line, 0, &buf[i], 0))
 408                         return(0);
 409                 goto descope;
 410         }
 411
 412         for (j = i; buf[i]; i++) {
 413                 if (' ' != buf[i])
 414                         continue;
 415
 416                 /* Escaped whitespace. */
 417                 if (i && ' ' == buf[i] && '\\' == buf[i - 1])
 418                         continue;
 419
 420                 buf[i++] = 0;
 421                 if ( ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 422                         return(0);
 423
 424                 for ( ; ' ' == buf[i]; i++)
 425                         /* Skip trailing whitespace. */ ;
 426
 427                 j = i;
 428                 if (0 == buf[i])
 429                         break;
 430         }
 431
 432         if (j != i && ! pstring(m, line, j, &buf[j], (size_t)(i - j)))
 433                 return(0);
 434
 435 descope:
 436
 437         /*
 438          * Co-ordinate what happens with having a next-line scope open:
 439          * first close out the element scope (if applicable), then close
 440          * out the block scope (also if applicable).
 441          */
 442
 443         if (MAN_ELINE & m->flags) {
 444                 m->flags &= ~MAN_ELINE;
 445                 if ( ! man_unscope(m, m->last->parent))
 446                         return(0);
 447         }
 448
 449         if ( ! (MAN_BLINE & m->flags))
 450                 return(1);
 451         m->flags &= ~MAN_BLINE;
 452
 453         if ( ! man_unscope(m, m->last->parent))
 454                 return(0);
 455         return(man_body_alloc(m, line, 0, m->last->tok));
 456 }
 457
 458
 459 static int
 460 macrowarn(struct man *m, int ln, const char *buf)
 461 {
 462         if ( ! (MAN_IGN_MACRO & m->pflags))
 463                 return(man_verr(m, ln, 0,
 464                                 "unknown macro: %s%s",
 465                                 buf, strlen(buf) > 3 ? "..." : ""));
 466         return(man_vwarn(m, ln, 0, "unknown macro: %s%s",
 467                                 buf, strlen(buf) > 3 ? "..." : ""));
 468 }
 469
 470
 471 int
 472 man_pmacro(struct man *m, int ln, char *buf)
 473 {
 474         int              i, j, c, ppos, fl;
 475         char             mac[5];
 476         struct man_node *n;
 477
 478         /* Comments and empties are quickly ignored. */
 479
 480         fl = m->flags;
 481
 482         if (0 == buf[1])
 483                 goto out;
 484
 485         i = 1;
 486
 487         if (' ' == buf[i]) {
 488                 i++;
 489                 while (buf[i] && ' ' == buf[i])
 490                         i++;
 491                 if (0 == buf[i])
 492                         goto out;
 493         }
 494
 495         ppos = i;
 496
 497         /* Copy the first word into a nil-terminated buffer. */
 498
 499         for (j = 0; j < 4; j++, i++) {
 500                 if (0 == (mac[j] = buf[i]))
 501                         break;
 502                 else if (' ' == buf[i])
 503                         break;
 504
 505                 /* Check for invalid characters. */
 506
 507                 if (isgraph((u_char)buf[i]))
 508                         continue;
 509                 return(man_perr(m, ln, i, WNPRINT));
 510         }
 511
 512         mac[j] = 0;
 513
 514         if (j == 4 || j < 1) {
 515                 if ( ! (MAN_IGN_MACRO & m->pflags)) {
 516                         (void)man_perr(m, ln, ppos, WMACROFORM);
 517                         goto err;
 518                 }
 519                 if ( ! man_pwarn(m, ln, ppos, WMACROFORM))
 520                         goto err;
 521                 return(1);
 522         }
 523
 524         if (MAN_MAX == (c = man_hash_find(mac))) {
 525                 if ( ! macrowarn(m, ln, mac))
 526                         goto err;
 527                 return(1);
 528         }
 529
 530         /* The macro is sane.  Jump to the next word. */
 531
 532         while (buf[i] && ' ' == buf[i])
 533                 i++;
 534
 535         /* Remove prior ELINE macro, if applicable. */
 536
 537         if (m->flags & MAN_ELINE) {
 538                 n = m->last;
 539                 assert(NULL == n->child);
 540                 assert(0 == n->nchild);
 541                 if ( ! man_nwarn(m, n, WLNSCOPE))
 542                         return(0);
 543
 544                 if (n->prev) {
 545                         assert(n != n->parent->child);
 546                         assert(n == n->prev->next);
 547                         n->prev->next = NULL;
 548                         m->last = n->prev;
 549                         m->next = MAN_NEXT_SIBLING;
 550                 } else {
 551                         assert(n == n->parent->child);
 552                         n->parent->child = NULL;
 553                         m->last = n->parent;
 554                         m->next = MAN_NEXT_CHILD;
 555                 }
 556
 557                 man_node_free(n);
 558                 m->flags &= ~MAN_ELINE;
 559         }
 560
 561         /* Begin recursive parse sequence. */
 562
 563         assert(man_macros[c].fp);
 564
 565         if ( ! (*man_macros[c].fp)(m, c, ln, ppos, &i, buf))
 566                 goto err;
 567
 568 out:
 569         if ( ! (MAN_BLINE & fl))
 570                 return(1);
 571
 572         /*
 573          * If we've opened a new next-line element scope, then return
 574          * now, as the next line will close out the block scope.
 575          */
 576
 577         if (MAN_ELINE & m->flags)
 578                 return(1);
 579
 580         /* Close out the block scope opened in the prior line.  */
 581
 582         assert(MAN_BLINE & m->flags);
 583         m->flags &= ~MAN_BLINE;
 584
 585         if ( ! man_unscope(m, m->last->parent))
 586                 return(0);
 587         return(man_body_alloc(m, ln, 0, m->last->tok));
 588
 589 err:    /* Error out. */
 590
 591         m->flags |= MAN_HALT;
 592         return(0);
 593 }
 594
 595
 596 int
 597 man_verr(struct man *man, int ln, int pos, const char *fmt, ...)
 598 {
 599         char             buf[256];
 600         va_list          ap;
 601
 602         if (NULL == man->cb.man_err)
 603                 return(0);
 604
 605         va_start(ap, fmt);
 606         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 607         va_end(ap);
 608         return((*man->cb.man_err)(man->data, ln, pos, buf));
 609 }
 610
 611
 612 int
 613 man_vwarn(struct man *man, int ln, int pos, const char *fmt, ...)
 614 {
 615         char             buf[256];
 616         va_list          ap;
 617
 618         if (NULL == man->cb.man_warn)
 619                 return(0);
 620
 621         va_start(ap, fmt);
 622         (void)vsnprintf(buf, sizeof(buf) - 1, fmt, ap);
 623         va_end(ap);
 624         return((*man->cb.man_warn)(man->data, ln, pos, buf));
 625 }
 626
 627
 628 int
 629 man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 630 {
 631         const char       *p;
 632
 633         p = __man_merrnames[(int)type];
 634         assert(p);
 635
 636         if (iserr)
 637                 return(man_verr(m, line, pos, p));
 638
 639         return(man_vwarn(m, line, pos, p));
 640 }