mdoc.c

   1 /*      $Id: mdoc.c,v 1.225 2014/09/06 23:24:32 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010, 2012, 2013, 2014 Ingo Schwarze <schwarze@openbsd.org>
   5  *
   6  * Permission to use, copy, modify, and distribute this software for any
   7  * purpose with or without fee is hereby granted, provided that the above
   8  * copyright notice and this permission notice appear in all copies.
   9  *
  10  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  11  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  12  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  13  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  14  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  15  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  16  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  17  */
  18 #include "config.h"
  19
  20 #include <sys/types.h>
  21
  22 #include <assert.h>
  23 #include <ctype.h>
  24 #include <stdarg.h>
  25 #include <stdio.h>
  26 #include <stdlib.h>
  27 #include <string.h>
  28 #include <time.h>
  29
  30 #include "mdoc.h"
  31 #include "mandoc.h"
  32 #include "mandoc_aux.h"
  33 #include "libmdoc.h"
  34 #include "libmandoc.h"
  35
  36 const   char *const __mdoc_macronames[MDOC_MAX + 1] = {
  37         "Ap",           "Dd",           "Dt",           "Os",
  38         "Sh",           "Ss",           "Pp",           "D1",
  39         "Dl",           "Bd",           "Ed",           "Bl",
  40         "El",           "It",           "Ad",           "An",
  41         "Ar",           "Cd",           "Cm",           "Dv",
  42         "Er",           "Ev",           "Ex",           "Fa",
  43         "Fd",           "Fl",           "Fn",           "Ft",
  44         "Ic",           "In",           "Li",           "Nd",
  45         "Nm",           "Op",           "Ot",           "Pa",
  46         "Rv",           "St",           "Va",           "Vt",
  47         "Xr",           "%A",           "%B",           "%D",
  48         "%I",           "%J",           "%N",           "%O",
  49         "%P",           "%R",           "%T",           "%V",
  50         "Ac",           "Ao",           "Aq",           "At",
  51         "Bc",           "Bf",           "Bo",           "Bq",
  52         "Bsx",          "Bx",           "Db",           "Dc",
  53         "Do",           "Dq",           "Ec",           "Ef",
  54         "Em",           "Eo",           "Fx",           "Ms",
  55         "No",           "Ns",           "Nx",           "Ox",
  56         "Pc",           "Pf",           "Po",           "Pq",
  57         "Qc",           "Ql",           "Qo",           "Qq",
  58         "Re",           "Rs",           "Sc",           "So",
  59         "Sq",           "Sm",           "Sx",           "Sy",
  60         "Tn",           "Ux",           "Xc",           "Xo",
  61         "Fo",           "Fc",           "Oo",           "Oc",
  62         "Bk",           "Ek",           "Bt",           "Hf",
  63         "Fr",           "Ud",           "Lb",           "Lp",
  64         "Lk",           "Mt",           "Brq",          "Bro",
  65         "Brc",          "%C",           "Es",           "En",
  66         "Dx",           "%Q",           "br",           "sp",
  67         "%U",           "Ta",           "ll",           "text",
  68         };
  69
  70 const   char *const __mdoc_argnames[MDOC_ARG_MAX] = {
  71         "split",                "nosplit",              "ragged",
  72         "unfilled",             "literal",              "file",
  73         "offset",               "bullet",               "dash",
  74         "hyphen",               "item",                 "enum",
  75         "tag",                  "diag",                 "hang",
  76         "ohang",                "inset",                "column",
  77         "width",                "compact",              "std",
  78         "filled",               "words",                "emphasis",
  79         "symbolic",             "nested",               "centered"
  80         };
  81
  82 const   char * const *mdoc_macronames = __mdoc_macronames;
  83 const   char * const *mdoc_argnames = __mdoc_argnames;
  84
  85 static  void              mdoc_node_free(struct mdoc_node *);
  86 static  void              mdoc_node_unlink(struct mdoc *,
  87                                 struct mdoc_node *);
  88 static  void              mdoc_free1(struct mdoc *);
  89 static  void              mdoc_alloc1(struct mdoc *);
  90 static  struct mdoc_node *node_alloc(struct mdoc *, int, int,
  91                                 enum mdoct, enum mdoc_type);
  92 static  int               node_append(struct mdoc *,
  93                                 struct mdoc_node *);
  94 #if 0
  95 static  int               mdoc_preptext(struct mdoc *, int, char *, int);
  96 #endif
  97 static  int               mdoc_ptext(struct mdoc *, int, char *, int);
  98 static  int               mdoc_pmacro(struct mdoc *, int, char *, int);
  99
 100
 101 const struct mdoc_node *
 102 mdoc_node(const struct mdoc *mdoc)
 103 {
 104
 105         return(mdoc->first);
 106 }
 107
 108 const struct mdoc_meta *
 109 mdoc_meta(const struct mdoc *mdoc)
 110 {
 111
 112         return(&mdoc->meta);
 113 }
 114
 115 /*
 116  * Frees volatile resources (parse tree, meta-data, fields).
 117  */
 118 static void
 119 mdoc_free1(struct mdoc *mdoc)
 120 {
 121
 122         if (mdoc->first)
 123                 mdoc_node_delete(mdoc, mdoc->first);
 124         free(mdoc->meta.msec);
 125         free(mdoc->meta.vol);
 126         free(mdoc->meta.arch);
 127         free(mdoc->meta.date);
 128         free(mdoc->meta.title);
 129         free(mdoc->meta.os);
 130         free(mdoc->meta.name);
 131 }
 132
 133 /*
 134  * Allocate all volatile resources (parse tree, meta-data, fields).
 135  */
 136 static void
 137 mdoc_alloc1(struct mdoc *mdoc)
 138 {
 139
 140         memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
 141         mdoc->flags = 0;
 142         mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
 143         mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
 144         mdoc->first = mdoc->last;
 145         mdoc->last->type = MDOC_ROOT;
 146         mdoc->last->tok = MDOC_MAX;
 147         mdoc->next = MDOC_NEXT_CHILD;
 148 }
 149
 150 /*
 151  * Free up volatile resources (see mdoc_free1()) then re-initialises the
 152  * data with mdoc_alloc1().  After invocation, parse data has been reset
 153  * and the parser is ready for re-invocation on a new tree; however,
 154  * cross-parse non-volatile data is kept intact.
 155  */
 156 void
 157 mdoc_reset(struct mdoc *mdoc)
 158 {
 159
 160         mdoc_free1(mdoc);
 161         mdoc_alloc1(mdoc);
 162 }
 163
 164 /*
 165  * Completely free up all volatile and non-volatile parse resources.
 166  * After invocation, the pointer is no longer usable.
 167  */
 168 void
 169 mdoc_free(struct mdoc *mdoc)
 170 {
 171
 172         mdoc_free1(mdoc);
 173         free(mdoc);
 174 }
 175
 176 /*
 177  * Allocate volatile and non-volatile parse resources.
 178  */
 179 struct mdoc *
 180 mdoc_alloc(struct roff *roff, struct mparse *parse,
 181         const char *defos, int quick)
 182 {
 183         struct mdoc     *p;
 184
 185         p = mandoc_calloc(1, sizeof(struct mdoc));
 186
 187         p->parse = parse;
 188         p->defos = defos;
 189         p->quick = quick;
 190         p->roff = roff;
 191
 192         mdoc_hash_init();
 193         mdoc_alloc1(p);
 194         return(p);
 195 }
 196
 197 int
 198 mdoc_endparse(struct mdoc *mdoc)
 199 {
 200
 201         return(mdoc_macroend(mdoc));
 202 }
 203
 204 int
 205 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
 206 {
 207         struct mdoc_node *n;
 208
 209         n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
 210         n->eqn = ep;
 211
 212         if ( ! node_append(mdoc, n))
 213                 return(0);
 214
 215         mdoc->next = MDOC_NEXT_SIBLING;
 216         return(1);
 217 }
 218
 219 int
 220 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
 221 {
 222         struct mdoc_node *n;
 223
 224         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
 225         n->span = sp;
 226
 227         if ( ! node_append(mdoc, n))
 228                 return(0);
 229
 230         mdoc->next = MDOC_NEXT_SIBLING;
 231         return(1);
 232 }
 233
 234 /*
 235  * Main parse routine.  Parses a single line -- really just hands off to
 236  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
 237  */
 238 int
 239 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
 240 {
 241
 242         mdoc->flags |= MDOC_NEWLINE;
 243
 244         /*
 245          * Let the roff nS register switch SYNOPSIS mode early,
 246          * such that the parser knows at all times
 247          * whether this mode is on or off.
 248          * Note that this mode is also switched by the Sh macro.
 249          */
 250         if (roff_getreg(mdoc->roff, "nS"))
 251                 mdoc->flags |= MDOC_SYNOPSIS;
 252         else
 253                 mdoc->flags &= ~MDOC_SYNOPSIS;
 254
 255         return(roff_getcontrol(mdoc->roff, buf, &offs) ?
 256             mdoc_pmacro(mdoc, ln, buf, offs) :
 257             mdoc_ptext(mdoc, ln, buf, offs));
 258 }
 259
 260 int
 261 mdoc_macro(MACRO_PROT_ARGS)
 262 {
 263         assert(tok < MDOC_MAX);
 264
 265         if (mdoc->flags & MDOC_PBODY) {
 266                 if (tok == MDOC_Dt) {
 267                         mandoc_vmsg(MANDOCERR_DT_LATE,
 268                             mdoc->parse, line, ppos,
 269                             "Dt %s", buf + *pos);
 270                         return(1);
 271                 }
 272         } else if ( ! (mdoc_macros[tok].flags & MDOC_PROLOGUE)) {
 273                 if (mdoc->meta.title == NULL) {
 274                         mandoc_vmsg(MANDOCERR_DT_NOTITLE,
 275                             mdoc->parse, line, ppos, "%s %s",
 276                             mdoc_macronames[tok], buf + *pos);
 277                         mdoc->meta.title = mandoc_strdup("UNTITLED");
 278                 }
 279                 if (NULL == mdoc->meta.vol)
 280                         mdoc->meta.vol = mandoc_strdup("LOCAL");
 281                 mdoc->flags |= MDOC_PBODY;
 282         }
 283
 284         return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
 285 }
 286
 287
 288 static int
 289 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 290 {
 291
 292         assert(mdoc->last);
 293         assert(mdoc->first);
 294         assert(MDOC_ROOT != p->type);
 295
 296         switch (mdoc->next) {
 297         case MDOC_NEXT_SIBLING:
 298                 mdoc->last->next = p;
 299                 p->prev = mdoc->last;
 300                 p->parent = mdoc->last->parent;
 301                 break;
 302         case MDOC_NEXT_CHILD:
 303                 mdoc->last->child = p;
 304                 p->parent = mdoc->last;
 305                 break;
 306         default:
 307                 abort();
 308                 /* NOTREACHED */
 309         }
 310
 311         p->parent->nchild++;
 312
 313         /*
 314          * Copy over the normalised-data pointer of our parent.  Not
 315          * everybody has one, but copying a null pointer is fine.
 316          */
 317
 318         switch (p->type) {
 319         case MDOC_BODY:
 320                 if (ENDBODY_NOT != p->end)
 321                         break;
 322                 /* FALLTHROUGH */
 323         case MDOC_TAIL:
 324                 /* FALLTHROUGH */
 325         case MDOC_HEAD:
 326                 p->norm = p->parent->norm;
 327                 break;
 328         default:
 329                 break;
 330         }
 331
 332         if ( ! mdoc_valid_pre(mdoc, p))
 333                 return(0);
 334
 335         switch (p->type) {
 336         case MDOC_HEAD:
 337                 assert(MDOC_BLOCK == p->parent->type);
 338                 p->parent->head = p;
 339                 break;
 340         case MDOC_TAIL:
 341                 assert(MDOC_BLOCK == p->parent->type);
 342                 p->parent->tail = p;
 343                 break;
 344         case MDOC_BODY:
 345                 if (p->end)
 346                         break;
 347                 assert(MDOC_BLOCK == p->parent->type);
 348                 p->parent->body = p;
 349                 break;
 350         default:
 351                 break;
 352         }
 353
 354         mdoc->last = p;
 355
 356         switch (p->type) {
 357         case MDOC_TBL:
 358                 /* FALLTHROUGH */
 359         case MDOC_TEXT:
 360                 if ( ! mdoc_valid_post(mdoc))
 361                         return(0);
 362                 break;
 363         default:
 364                 break;
 365         }
 366
 367         return(1);
 368 }
 369
 370 static struct mdoc_node *
 371 node_alloc(struct mdoc *mdoc, int line, int pos,
 372                 enum mdoct tok, enum mdoc_type type)
 373 {
 374         struct mdoc_node *p;
 375
 376         p = mandoc_calloc(1, sizeof(struct mdoc_node));
 377         p->sec = mdoc->lastsec;
 378         p->line = line;
 379         p->pos = pos;
 380         p->lastline = line;
 381         p->tok = tok;
 382         p->type = type;
 383
 384         /* Flag analysis. */
 385
 386         if (MDOC_SYNOPSIS & mdoc->flags)
 387                 p->flags |= MDOC_SYNPRETTY;
 388         else
 389                 p->flags &= ~MDOC_SYNPRETTY;
 390         if (MDOC_NEWLINE & mdoc->flags)
 391                 p->flags |= MDOC_LINE;
 392         mdoc->flags &= ~MDOC_NEWLINE;
 393
 394         return(p);
 395 }
 396
 397 int
 398 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 399 {
 400         struct mdoc_node *p;
 401
 402         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 403         if ( ! node_append(mdoc, p))
 404                 return(0);
 405         mdoc->next = MDOC_NEXT_CHILD;
 406         return(1);
 407 }
 408
 409 int
 410 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 411 {
 412         struct mdoc_node *p;
 413
 414         assert(mdoc->first);
 415         assert(mdoc->last);
 416
 417         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 418         if ( ! node_append(mdoc, p))
 419                 return(0);
 420         mdoc->next = MDOC_NEXT_CHILD;
 421         return(1);
 422 }
 423
 424 int
 425 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 426 {
 427         struct mdoc_node *p;
 428
 429         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 430         if ( ! node_append(mdoc, p))
 431                 return(0);
 432         mdoc->next = MDOC_NEXT_CHILD;
 433         return(1);
 434 }
 435
 436 int
 437 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 438                 struct mdoc_node *body, enum mdoc_endbody end)
 439 {
 440         struct mdoc_node *p;
 441
 442         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 443         p->pending = body;
 444         p->norm = body->norm;
 445         p->end = end;
 446         if ( ! node_append(mdoc, p))
 447                 return(0);
 448         mdoc->next = MDOC_NEXT_SIBLING;
 449         return(1);
 450 }
 451
 452 int
 453 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 454                 enum mdoct tok, struct mdoc_arg *args)
 455 {
 456         struct mdoc_node *p;
 457
 458         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 459         p->args = args;
 460         if (p->args)
 461                 (args->refcnt)++;
 462
 463         switch (tok) {
 464         case MDOC_Bd:
 465                 /* FALLTHROUGH */
 466         case MDOC_Bf:
 467                 /* FALLTHROUGH */
 468         case MDOC_Bl:
 469                 /* FALLTHROUGH */
 470         case MDOC_En:
 471                 /* FALLTHROUGH */
 472         case MDOC_Rs:
 473                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 474                 break;
 475         default:
 476                 break;
 477         }
 478
 479         if ( ! node_append(mdoc, p))
 480                 return(0);
 481         mdoc->next = MDOC_NEXT_CHILD;
 482         return(1);
 483 }
 484
 485 int
 486 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 487                 enum mdoct tok, struct mdoc_arg *args)
 488 {
 489         struct mdoc_node *p;
 490
 491         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 492         p->args = args;
 493         if (p->args)
 494                 (args->refcnt)++;
 495
 496         switch (tok) {
 497         case MDOC_An:
 498                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 499                 break;
 500         default:
 501                 break;
 502         }
 503
 504         if ( ! node_append(mdoc, p))
 505                 return(0);
 506         mdoc->next = MDOC_NEXT_CHILD;
 507         return(1);
 508 }
 509
 510 int
 511 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 512 {
 513         struct mdoc_node *n;
 514
 515         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 516         n->string = roff_strdup(mdoc->roff, p);
 517
 518         if ( ! node_append(mdoc, n))
 519                 return(0);
 520
 521         mdoc->next = MDOC_NEXT_SIBLING;
 522         return(1);
 523 }
 524
 525 void
 526 mdoc_word_append(struct mdoc *mdoc, const char *p)
 527 {
 528         struct mdoc_node        *n;
 529         char                    *addstr, *newstr;
 530
 531         n = mdoc->last;
 532         addstr = roff_strdup(mdoc->roff, p);
 533         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 534         free(addstr);
 535         free(n->string);
 536         n->string = newstr;
 537         mdoc->next = MDOC_NEXT_SIBLING;
 538 }
 539
 540 static void
 541 mdoc_node_free(struct mdoc_node *p)
 542 {
 543
 544         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 545                 free(p->norm);
 546         if (p->string)
 547                 free(p->string);
 548         if (p->args)
 549                 mdoc_argv_free(p->args);
 550         free(p);
 551 }
 552
 553 static void
 554 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 555 {
 556
 557         /* Adjust siblings. */
 558
 559         if (n->prev)
 560                 n->prev->next = n->next;
 561         if (n->next)
 562                 n->next->prev = n->prev;
 563
 564         /* Adjust parent. */
 565
 566         if (n->parent) {
 567                 n->parent->nchild--;
 568                 if (n->parent->child == n)
 569                         n->parent->child = n->prev ? n->prev : n->next;
 570                 if (n->parent->last == n)
 571                         n->parent->last = n->prev ? n->prev : NULL;
 572         }
 573
 574         /* Adjust parse point, if applicable. */
 575
 576         if (mdoc && mdoc->last == n) {
 577                 if (n->prev) {
 578                         mdoc->last = n->prev;
 579                         mdoc->next = MDOC_NEXT_SIBLING;
 580                 } else {
 581                         mdoc->last = n->parent;
 582                         mdoc->next = MDOC_NEXT_CHILD;
 583                 }
 584         }
 585
 586         if (mdoc && mdoc->first == n)
 587                 mdoc->first = NULL;
 588 }
 589
 590 void
 591 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 592 {
 593
 594         while (p->child) {
 595                 assert(p->nchild);
 596                 mdoc_node_delete(mdoc, p->child);
 597         }
 598         assert(0 == p->nchild);
 599
 600         mdoc_node_unlink(mdoc, p);
 601         mdoc_node_free(p);
 602 }
 603
 604 int
 605 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 606 {
 607
 608         mdoc_node_unlink(mdoc, p);
 609         return(node_append(mdoc, p));
 610 }
 611
 612 #if 0
 613 /*
 614  * Pre-treat a text line.
 615  * Text lines can consist of equations, which must be handled apart from
 616  * the regular text.
 617  * Thus, use this function to step through a line checking if it has any
 618  * equations embedded in it.
 619  * This must handle multiple equations AND equations that do not end at
 620  * the end-of-line, i.e., will re-enter in the next roff parse.
 621  */
 622 static int
 623 mdoc_preptext(struct mdoc *mdoc, int line, char *buf, int offs)
 624 {
 625         char            *start, *end;
 626         char             delim;
 627
 628         while ('\0' != buf[offs]) {
 629                 /* Mark starting position if eqn is set. */
 630                 start = NULL;
 631                 if ('\0' != (delim = roff_eqndelim(mdoc->roff)))
 632                         if (NULL != (start = strchr(buf + offs, delim)))
 633                                 *start++ = '\0';
 634
 635                 /* Parse text as normal. */
 636                 if ( ! mdoc_ptext(mdoc, line, buf, offs))
 637                         return(0);
 638
 639                 /* Continue only if an equation exists. */
 640                 if (NULL == start)
 641                         break;
 642
 643                 /* Read past the end of the equation. */
 644                 offs += start - (buf + offs);
 645                 assert(start == &buf[offs]);
 646                 if (NULL != (end = strchr(buf + offs, delim))) {
 647                         *end++ = '\0';
 648                         while (' ' == *end)
 649                                 end++;
 650                 }
 651
 652                 /* Parse the equation itself. */
 653                 roff_openeqn(mdoc->roff, NULL, line, offs, buf);
 654
 655                 /* Process a finished equation? */
 656                 if (roff_closeeqn(mdoc->roff))
 657                         if ( ! mdoc_addeqn(mdoc, roff_eqn(mdoc->roff)))
 658                                 return(0);
 659                 offs += (end - (buf + offs));
 660         }
 661
 662         return(1);
 663 }
 664 #endif
 665
 666 /*
 667  * Parse free-form text, that is, a line that does not begin with the
 668  * control character.
 669  */
 670 static int
 671 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 672 {
 673         char             *c, *ws, *end;
 674         struct mdoc_node *n;
 675
 676         assert(mdoc->last);
 677         n = mdoc->last;
 678
 679         /*
 680          * Divert directly to list processing if we're encountering a
 681          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 682          * (a MDOC_BODY means it's already open, in which case we should
 683          * process within its context in the normal way).
 684          */
 685
 686         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 687             LIST_column == n->norm->Bl.type) {
 688                 /* `Bl' is open without any children. */
 689                 mdoc->flags |= MDOC_FREECOL;
 690                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 691         }
 692
 693         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 694             NULL != n->parent &&
 695             MDOC_Bl == n->parent->tok &&
 696             LIST_column == n->parent->norm->Bl.type) {
 697                 /* `Bl' has block-level `It' children. */
 698                 mdoc->flags |= MDOC_FREECOL;
 699                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 700         }
 701
 702         /*
 703          * Search for the beginning of unescaped trailing whitespace (ws)
 704          * and for the first character not to be output (end).
 705          */
 706
 707         /* FIXME: replace with strcspn(). */
 708         ws = NULL;
 709         for (c = end = buf + offs; *c; c++) {
 710                 switch (*c) {
 711                 case ' ':
 712                         if (NULL == ws)
 713                                 ws = c;
 714                         continue;
 715                 case '\t':
 716                         /*
 717                          * Always warn about trailing tabs,
 718                          * even outside literal context,
 719                          * where they should be put on the next line.
 720                          */
 721                         if (NULL == ws)
 722                                 ws = c;
 723                         /*
 724                          * Strip trailing tabs in literal context only;
 725                          * outside, they affect the next line.
 726                          */
 727                         if (MDOC_LITERAL & mdoc->flags)
 728                                 continue;
 729                         break;
 730                 case '\\':
 731                         /* Skip the escaped character, too, if any. */
 732                         if (c[1])
 733                                 c++;
 734                         /* FALLTHROUGH */
 735                 default:
 736                         ws = NULL;
 737                         break;
 738                 }
 739                 end = c + 1;
 740         }
 741         *end = '\0';
 742
 743         if (ws)
 744                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 745                     line, (int)(ws-buf), NULL);
 746
 747         if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
 748                 mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
 749                     line, (int)(c - buf), NULL);
 750
 751                 /*
 752                  * Insert a `sp' in the case of a blank line.  Technically,
 753                  * blank lines aren't allowed, but enough manuals assume this
 754                  * behaviour that we want to work around it.
 755                  */
 756                 if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
 757                         return(0);
 758
 759                 mdoc->next = MDOC_NEXT_SIBLING;
 760
 761                 return(mdoc_valid_post(mdoc));
 762         }
 763
 764         if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
 765                 return(0);
 766
 767         if (MDOC_LITERAL & mdoc->flags)
 768                 return(1);
 769
 770         /*
 771          * End-of-sentence check.  If the last character is an unescaped
 772          * EOS character, then flag the node as being the end of a
 773          * sentence.  The front-end will know how to interpret this.
 774          */
 775
 776         assert(buf < end);
 777
 778         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
 779                 mdoc->last->flags |= MDOC_EOS;
 780
 781         return(1);
 782 }
 783
 784 /*
 785  * Parse a macro line, that is, a line beginning with the control
 786  * character.
 787  */
 788 static int
 789 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 790 {
 791         enum mdoct        tok;
 792         int               i, sv;
 793         char              mac[5];
 794         struct mdoc_node *n;
 795
 796         sv = offs;
 797
 798         /*
 799          * Copy the first word into a nil-terminated buffer.
 800          * Stop copying when a tab, space, or eoln is encountered.
 801          */
 802
 803         i = 0;
 804         while (i < 4 && '\0' != buf[offs] && ' ' != buf[offs] &&
 805             '\t' != buf[offs])
 806                 mac[i++] = buf[offs++];
 807
 808         mac[i] = '\0';
 809
 810         tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 811
 812         if (MDOC_MAX == tok) {
 813                 mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
 814                     ln, sv, buf + sv - 1);
 815                 return(1);
 816         }
 817
 818         /* Disregard the first trailing tab, if applicable. */
 819
 820         if ('\t' == buf[offs])
 821                 offs++;
 822
 823         /* Jump to the next non-whitespace word. */
 824
 825         while (buf[offs] && ' ' == buf[offs])
 826                 offs++;
 827
 828         /*
 829          * Trailing whitespace.  Note that tabs are allowed to be passed
 830          * into the parser as "text", so we only warn about spaces here.
 831          */
 832
 833         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 834                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 835                     ln, offs - 1, NULL);
 836
 837         /*
 838          * If an initial macro or a list invocation, divert directly
 839          * into macro processing.
 840          */
 841
 842         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok)
 843                 return(mdoc_macro(mdoc, tok, ln, sv, &offs, buf));
 844
 845         n = mdoc->last;
 846         assert(mdoc->last);
 847
 848         /*
 849          * If the first macro of a `Bl -column', open an `It' block
 850          * context around the parsed macro.
 851          */
 852
 853         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 854             LIST_column == n->norm->Bl.type) {
 855                 mdoc->flags |= MDOC_FREECOL;
 856                 return(mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf));
 857         }
 858
 859         /*
 860          * If we're following a block-level `It' within a `Bl -column'
 861          * context (perhaps opened in the above block or in ptext()),
 862          * then open an `It' block context around the parsed macro.
 863          */
 864
 865         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 866             NULL != n->parent &&
 867             MDOC_Bl == n->parent->tok &&
 868             LIST_column == n->parent->norm->Bl.type) {
 869                 mdoc->flags |= MDOC_FREECOL;
 870                 return(mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf));
 871         }
 872
 873         /* Normal processing of a macro. */
 874
 875         if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 876                 return(0);
 877
 878         /* In quick mode (for mandocdb), abort after the NAME section. */
 879
 880         if (mdoc->quick && MDOC_Sh == tok &&
 881             SEC_NAME != mdoc->last->sec)
 882                 return(2);
 883
 884         return(1);
 885 }
 886
 887 enum mdelim
 888 mdoc_isdelim(const char *p)
 889 {
 890
 891         if ('\0' == p[0])
 892                 return(DELIM_NONE);
 893
 894         if ('\0' == p[1])
 895                 switch (p[0]) {
 896                 case '(':
 897                         /* FALLTHROUGH */
 898                 case '[':
 899                         return(DELIM_OPEN);
 900                 case '|':
 901                         return(DELIM_MIDDLE);
 902                 case '.':
 903                         /* FALLTHROUGH */
 904                 case ',':
 905                         /* FALLTHROUGH */
 906                 case ';':
 907                         /* FALLTHROUGH */
 908                 case ':':
 909                         /* FALLTHROUGH */
 910                 case '?':
 911                         /* FALLTHROUGH */
 912                 case '!':
 913                         /* FALLTHROUGH */
 914                 case ')':
 915                         /* FALLTHROUGH */
 916                 case ']':
 917                         return(DELIM_CLOSE);
 918                 default:
 919                         return(DELIM_NONE);
 920                 }
 921
 922         if ('\\' != p[0])
 923                 return(DELIM_NONE);
 924
 925         if (0 == strcmp(p + 1, "."))
 926                 return(DELIM_CLOSE);
 927         if (0 == strcmp(p + 1, "fR|\\fP"))
 928                 return(DELIM_MIDDLE);
 929
 930         return(DELIM_NONE);
 931 }
 932
 933 void
 934 mdoc_deroff(char **dest, const struct mdoc_node *n)
 935 {
 936         char    *cp;
 937         size_t   sz;
 938
 939         if (MDOC_TEXT != n->type) {
 940                 for (n = n->child; n; n = n->next)
 941                         mdoc_deroff(dest, n);
 942                 return;
 943         }
 944
 945         /* Skip leading whitespace. */
 946
 947         for (cp = n->string; '\0' != *cp; cp++)
 948                 if (0 == isspace((unsigned char)*cp))
 949                         break;
 950
 951         /* Skip trailing whitespace. */
 952
 953         for (sz = strlen(cp); sz; sz--)
 954                 if (0 == isspace((unsigned char)cp[sz-1]))
 955                         break;
 956
 957         /* Skip empty strings. */
 958
 959         if (0 == sz)
 960                 return;
 961
 962         if (NULL == *dest) {
 963                 *dest = mandoc_strndup(cp, sz);
 964                 return;
 965         }
 966
 967         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 968         free(*dest);
 969         *dest = cp;
 970 }