mdoc.c

   1 /*      $Id: mdoc.c,v 1.238 2015/02/12 13:00:52 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2010, 2012-2015 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  void              node_append(struct mdoc *, struct mdoc_node *);
  93 static  int               mdoc_ptext(struct mdoc *, int, char *, int);
  94 static  int               mdoc_pmacro(struct mdoc *, int, char *, int);
  95
  96
  97 const struct mdoc_node *
  98 mdoc_node(const struct mdoc *mdoc)
  99 {
 100
 101         return(mdoc->first);
 102 }
 103
 104 const struct mdoc_meta *
 105 mdoc_meta(const struct mdoc *mdoc)
 106 {
 107
 108         return(&mdoc->meta);
 109 }
 110
 111 /*
 112  * Frees volatile resources (parse tree, meta-data, fields).
 113  */
 114 static void
 115 mdoc_free1(struct mdoc *mdoc)
 116 {
 117
 118         if (mdoc->first)
 119                 mdoc_node_delete(mdoc, mdoc->first);
 120         free(mdoc->meta.msec);
 121         free(mdoc->meta.vol);
 122         free(mdoc->meta.arch);
 123         free(mdoc->meta.date);
 124         free(mdoc->meta.title);
 125         free(mdoc->meta.os);
 126         free(mdoc->meta.name);
 127 }
 128
 129 /*
 130  * Allocate all volatile resources (parse tree, meta-data, fields).
 131  */
 132 static void
 133 mdoc_alloc1(struct mdoc *mdoc)
 134 {
 135
 136         memset(&mdoc->meta, 0, sizeof(struct mdoc_meta));
 137         mdoc->flags = 0;
 138         mdoc->lastnamed = mdoc->lastsec = SEC_NONE;
 139         mdoc->last = mandoc_calloc(1, sizeof(struct mdoc_node));
 140         mdoc->first = mdoc->last;
 141         mdoc->last->type = MDOC_ROOT;
 142         mdoc->last->tok = MDOC_MAX;
 143         mdoc->next = MDOC_NEXT_CHILD;
 144 }
 145
 146 /*
 147  * Free up volatile resources (see mdoc_free1()) then re-initialises the
 148  * data with mdoc_alloc1().  After invocation, parse data has been reset
 149  * and the parser is ready for re-invocation on a new tree; however,
 150  * cross-parse non-volatile data is kept intact.
 151  */
 152 void
 153 mdoc_reset(struct mdoc *mdoc)
 154 {
 155
 156         mdoc_free1(mdoc);
 157         mdoc_alloc1(mdoc);
 158 }
 159
 160 /*
 161  * Completely free up all volatile and non-volatile parse resources.
 162  * After invocation, the pointer is no longer usable.
 163  */
 164 void
 165 mdoc_free(struct mdoc *mdoc)
 166 {
 167
 168         mdoc_free1(mdoc);
 169         free(mdoc);
 170 }
 171
 172 /*
 173  * Allocate volatile and non-volatile parse resources.
 174  */
 175 struct mdoc *
 176 mdoc_alloc(struct roff *roff, struct mparse *parse,
 177         const char *defos, int quick)
 178 {
 179         struct mdoc     *p;
 180
 181         p = mandoc_calloc(1, sizeof(struct mdoc));
 182
 183         p->parse = parse;
 184         p->defos = defos;
 185         p->quick = quick;
 186         p->roff = roff;
 187
 188         mdoc_hash_init();
 189         mdoc_alloc1(p);
 190         return(p);
 191 }
 192
 193 void
 194 mdoc_endparse(struct mdoc *mdoc)
 195 {
 196
 197         mdoc_macroend(mdoc);
 198 }
 199
 200 void
 201 mdoc_addeqn(struct mdoc *mdoc, const struct eqn *ep)
 202 {
 203         struct mdoc_node *n;
 204
 205         n = node_alloc(mdoc, ep->ln, ep->pos, MDOC_MAX, MDOC_EQN);
 206         n->eqn = ep;
 207         if (ep->ln > mdoc->last->line)
 208                 n->flags |= MDOC_LINE;
 209         node_append(mdoc, n);
 210         mdoc->next = MDOC_NEXT_SIBLING;
 211 }
 212
 213 void
 214 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
 215 {
 216         struct mdoc_node *n;
 217
 218         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
 219         n->span = sp;
 220         node_append(mdoc, n);
 221         mdoc->next = MDOC_NEXT_SIBLING;
 222 }
 223
 224 /*
 225  * Main parse routine.  Parses a single line -- really just hands off to
 226  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
 227  */
 228 int
 229 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
 230 {
 231
 232         if (mdoc->last->type != MDOC_EQN || ln > mdoc->last->line)
 233                 mdoc->flags |= MDOC_NEWLINE;
 234
 235         /*
 236          * Let the roff nS register switch SYNOPSIS mode early,
 237          * such that the parser knows at all times
 238          * whether this mode is on or off.
 239          * Note that this mode is also switched by the Sh macro.
 240          */
 241         if (roff_getreg(mdoc->roff, "nS"))
 242                 mdoc->flags |= MDOC_SYNOPSIS;
 243         else
 244                 mdoc->flags &= ~MDOC_SYNOPSIS;
 245
 246         return(roff_getcontrol(mdoc->roff, buf, &offs) ?
 247             mdoc_pmacro(mdoc, ln, buf, offs) :
 248             mdoc_ptext(mdoc, ln, buf, offs));
 249 }
 250
 251 void
 252 mdoc_macro(MACRO_PROT_ARGS)
 253 {
 254         assert(tok < MDOC_MAX);
 255
 256         if (mdoc->flags & MDOC_PBODY) {
 257                 if (tok == MDOC_Dt) {
 258                         mandoc_vmsg(MANDOCERR_DT_LATE,
 259                             mdoc->parse, line, ppos,
 260                             "Dt %s", buf + *pos);
 261                         return;
 262                 }
 263         } else if ( ! (mdoc_macros[tok].flags & MDOC_PROLOGUE)) {
 264                 if (mdoc->meta.title == NULL) {
 265                         mandoc_vmsg(MANDOCERR_DT_NOTITLE,
 266                             mdoc->parse, line, ppos, "%s %s",
 267                             mdoc_macronames[tok], buf + *pos);
 268                         mdoc->meta.title = mandoc_strdup("UNTITLED");
 269                 }
 270                 if (NULL == mdoc->meta.vol)
 271                         mdoc->meta.vol = mandoc_strdup("LOCAL");
 272                 mdoc->flags |= MDOC_PBODY;
 273         }
 274         (*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf);
 275 }
 276
 277
 278 static void
 279 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 280 {
 281
 282         assert(mdoc->last);
 283         assert(mdoc->first);
 284         assert(MDOC_ROOT != p->type);
 285
 286         switch (mdoc->next) {
 287         case MDOC_NEXT_SIBLING:
 288                 mdoc->last->next = p;
 289                 p->prev = mdoc->last;
 290                 p->parent = mdoc->last->parent;
 291                 break;
 292         case MDOC_NEXT_CHILD:
 293                 mdoc->last->child = p;
 294                 p->parent = mdoc->last;
 295                 break;
 296         default:
 297                 abort();
 298                 /* NOTREACHED */
 299         }
 300
 301         p->parent->nchild++;
 302
 303         /*
 304          * Copy over the normalised-data pointer of our parent.  Not
 305          * everybody has one, but copying a null pointer is fine.
 306          */
 307
 308         switch (p->type) {
 309         case MDOC_BODY:
 310                 if (ENDBODY_NOT != p->end)
 311                         break;
 312                 /* FALLTHROUGH */
 313         case MDOC_TAIL:
 314                 /* FALLTHROUGH */
 315         case MDOC_HEAD:
 316                 p->norm = p->parent->norm;
 317                 break;
 318         default:
 319                 break;
 320         }
 321
 322         mdoc_valid_pre(mdoc, p);
 323
 324         switch (p->type) {
 325         case MDOC_HEAD:
 326                 assert(MDOC_BLOCK == p->parent->type);
 327                 p->parent->head = p;
 328                 break;
 329         case MDOC_TAIL:
 330                 assert(MDOC_BLOCK == p->parent->type);
 331                 p->parent->tail = p;
 332                 break;
 333         case MDOC_BODY:
 334                 if (p->end)
 335                         break;
 336                 assert(MDOC_BLOCK == p->parent->type);
 337                 p->parent->body = p;
 338                 break;
 339         default:
 340                 break;
 341         }
 342
 343         mdoc->last = p;
 344
 345         switch (p->type) {
 346         case MDOC_TBL:
 347                 /* FALLTHROUGH */
 348         case MDOC_TEXT:
 349                 mdoc_valid_post(mdoc);
 350                 break;
 351         default:
 352                 break;
 353         }
 354 }
 355
 356 static struct mdoc_node *
 357 node_alloc(struct mdoc *mdoc, int line, int pos,
 358                 enum mdoct tok, enum mdoc_type type)
 359 {
 360         struct mdoc_node *p;
 361
 362         p = mandoc_calloc(1, sizeof(struct mdoc_node));
 363         p->sec = mdoc->lastsec;
 364         p->line = line;
 365         p->pos = pos;
 366         p->tok = tok;
 367         p->type = type;
 368
 369         /* Flag analysis. */
 370
 371         if (MDOC_SYNOPSIS & mdoc->flags)
 372                 p->flags |= MDOC_SYNPRETTY;
 373         else
 374                 p->flags &= ~MDOC_SYNPRETTY;
 375         if (MDOC_NEWLINE & mdoc->flags)
 376                 p->flags |= MDOC_LINE;
 377         mdoc->flags &= ~MDOC_NEWLINE;
 378
 379         return(p);
 380 }
 381
 382 void
 383 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 384 {
 385         struct mdoc_node *p;
 386
 387         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 388         node_append(mdoc, p);
 389         mdoc->next = MDOC_NEXT_CHILD;
 390 }
 391
 392 struct mdoc_node *
 393 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 394 {
 395         struct mdoc_node *p;
 396
 397         assert(mdoc->first);
 398         assert(mdoc->last);
 399         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 400         node_append(mdoc, p);
 401         mdoc->next = MDOC_NEXT_CHILD;
 402         return(p);
 403 }
 404
 405 struct mdoc_node *
 406 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 407 {
 408         struct mdoc_node *p;
 409
 410         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 411         node_append(mdoc, p);
 412         mdoc->next = MDOC_NEXT_CHILD;
 413         return(p);
 414 }
 415
 416 struct mdoc_node *
 417 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 418                 struct mdoc_node *body, enum mdoc_endbody end)
 419 {
 420         struct mdoc_node *p;
 421
 422         body->flags |= MDOC_ENDED;
 423         body->parent->flags |= MDOC_ENDED;
 424         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 425         p->body = body;
 426         p->norm = body->norm;
 427         p->end = end;
 428         node_append(mdoc, p);
 429         mdoc->next = MDOC_NEXT_SIBLING;
 430         return(p);
 431 }
 432
 433 struct mdoc_node *
 434 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 435                 enum mdoct tok, struct mdoc_arg *args)
 436 {
 437         struct mdoc_node *p;
 438
 439         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 440         p->args = args;
 441         if (p->args)
 442                 (args->refcnt)++;
 443
 444         switch (tok) {
 445         case MDOC_Bd:
 446                 /* FALLTHROUGH */
 447         case MDOC_Bf:
 448                 /* FALLTHROUGH */
 449         case MDOC_Bl:
 450                 /* FALLTHROUGH */
 451         case MDOC_En:
 452                 /* FALLTHROUGH */
 453         case MDOC_Rs:
 454                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 455                 break;
 456         default:
 457                 break;
 458         }
 459         node_append(mdoc, p);
 460         mdoc->next = MDOC_NEXT_CHILD;
 461         return(p);
 462 }
 463
 464 void
 465 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 466                 enum mdoct tok, struct mdoc_arg *args)
 467 {
 468         struct mdoc_node *p;
 469
 470         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 471         p->args = args;
 472         if (p->args)
 473                 (args->refcnt)++;
 474
 475         switch (tok) {
 476         case MDOC_An:
 477                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 478                 break;
 479         default:
 480                 break;
 481         }
 482         node_append(mdoc, p);
 483         mdoc->next = MDOC_NEXT_CHILD;
 484 }
 485
 486 void
 487 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 488 {
 489         struct mdoc_node *n;
 490
 491         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 492         n->string = roff_strdup(mdoc->roff, p);
 493         node_append(mdoc, n);
 494         mdoc->next = MDOC_NEXT_SIBLING;
 495 }
 496
 497 void
 498 mdoc_word_append(struct mdoc *mdoc, const char *p)
 499 {
 500         struct mdoc_node        *n;
 501         char                    *addstr, *newstr;
 502
 503         n = mdoc->last;
 504         addstr = roff_strdup(mdoc->roff, p);
 505         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 506         free(addstr);
 507         free(n->string);
 508         n->string = newstr;
 509         mdoc->next = MDOC_NEXT_SIBLING;
 510 }
 511
 512 static void
 513 mdoc_node_free(struct mdoc_node *p)
 514 {
 515
 516         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 517                 free(p->norm);
 518         if (p->string)
 519                 free(p->string);
 520         if (p->args)
 521                 mdoc_argv_free(p->args);
 522         free(p);
 523 }
 524
 525 static void
 526 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 527 {
 528
 529         /* Adjust siblings. */
 530
 531         if (n->prev)
 532                 n->prev->next = n->next;
 533         if (n->next)
 534                 n->next->prev = n->prev;
 535
 536         /* Adjust parent. */
 537
 538         if (n->parent) {
 539                 n->parent->nchild--;
 540                 if (n->parent->child == n)
 541                         n->parent->child = n->prev ? n->prev : n->next;
 542                 if (n->parent->last == n)
 543                         n->parent->last = n->prev ? n->prev : NULL;
 544         }
 545
 546         /* Adjust parse point, if applicable. */
 547
 548         if (mdoc && mdoc->last == n) {
 549                 if (n->prev) {
 550                         mdoc->last = n->prev;
 551                         mdoc->next = MDOC_NEXT_SIBLING;
 552                 } else {
 553                         mdoc->last = n->parent;
 554                         mdoc->next = MDOC_NEXT_CHILD;
 555                 }
 556         }
 557
 558         if (mdoc && mdoc->first == n)
 559                 mdoc->first = NULL;
 560 }
 561
 562 void
 563 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 564 {
 565
 566         while (p->child) {
 567                 assert(p->nchild);
 568                 mdoc_node_delete(mdoc, p->child);
 569         }
 570         assert(0 == p->nchild);
 571
 572         mdoc_node_unlink(mdoc, p);
 573         mdoc_node_free(p);
 574 }
 575
 576 void
 577 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 578 {
 579
 580         mdoc_node_unlink(mdoc, p);
 581         node_append(mdoc, p);
 582 }
 583
 584 /*
 585  * Parse free-form text, that is, a line that does not begin with the
 586  * control character.
 587  */
 588 static int
 589 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 590 {
 591         char             *c, *ws, *end;
 592         struct mdoc_node *n;
 593
 594         assert(mdoc->last);
 595         n = mdoc->last;
 596
 597         /*
 598          * Divert directly to list processing if we're encountering a
 599          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 600          * (a MDOC_BODY means it's already open, in which case we should
 601          * process within its context in the normal way).
 602          */
 603
 604         if (n->tok == MDOC_Bl && n->type == MDOC_BODY &&
 605             n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) {
 606                 /* `Bl' is open without any children. */
 607                 mdoc->flags |= MDOC_FREECOL;
 608                 mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
 609                 return(1);
 610         }
 611
 612         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 613             NULL != n->parent &&
 614             MDOC_Bl == n->parent->tok &&
 615             LIST_column == n->parent->norm->Bl.type) {
 616                 /* `Bl' has block-level `It' children. */
 617                 mdoc->flags |= MDOC_FREECOL;
 618                 mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf);
 619                 return(1);
 620         }
 621
 622         /*
 623          * Search for the beginning of unescaped trailing whitespace (ws)
 624          * and for the first character not to be output (end).
 625          */
 626
 627         /* FIXME: replace with strcspn(). */
 628         ws = NULL;
 629         for (c = end = buf + offs; *c; c++) {
 630                 switch (*c) {
 631                 case ' ':
 632                         if (NULL == ws)
 633                                 ws = c;
 634                         continue;
 635                 case '\t':
 636                         /*
 637                          * Always warn about trailing tabs,
 638                          * even outside literal context,
 639                          * where they should be put on the next line.
 640                          */
 641                         if (NULL == ws)
 642                                 ws = c;
 643                         /*
 644                          * Strip trailing tabs in literal context only;
 645                          * outside, they affect the next line.
 646                          */
 647                         if (MDOC_LITERAL & mdoc->flags)
 648                                 continue;
 649                         break;
 650                 case '\\':
 651                         /* Skip the escaped character, too, if any. */
 652                         if (c[1])
 653                                 c++;
 654                         /* FALLTHROUGH */
 655                 default:
 656                         ws = NULL;
 657                         break;
 658                 }
 659                 end = c + 1;
 660         }
 661         *end = '\0';
 662
 663         if (ws)
 664                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 665                     line, (int)(ws-buf), NULL);
 666
 667         if (buf[offs] == '\0' && ! (mdoc->flags & MDOC_LITERAL)) {
 668                 mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
 669                     line, (int)(c - buf), NULL);
 670
 671                 /*
 672                  * Insert a `sp' in the case of a blank line.  Technically,
 673                  * blank lines aren't allowed, but enough manuals assume this
 674                  * behaviour that we want to work around it.
 675                  */
 676                 mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL);
 677                 mdoc->next = MDOC_NEXT_SIBLING;
 678                 mdoc_valid_post(mdoc);
 679                 return(1);
 680         }
 681
 682         mdoc_word_alloc(mdoc, line, offs, buf+offs);
 683
 684         if (mdoc->flags & MDOC_LITERAL)
 685                 return(1);
 686
 687         /*
 688          * End-of-sentence check.  If the last character is an unescaped
 689          * EOS character, then flag the node as being the end of a
 690          * sentence.  The front-end will know how to interpret this.
 691          */
 692
 693         assert(buf < end);
 694
 695         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
 696                 mdoc->last->flags |= MDOC_EOS;
 697         return(1);
 698 }
 699
 700 /*
 701  * Parse a macro line, that is, a line beginning with the control
 702  * character.
 703  */
 704 static int
 705 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 706 {
 707         struct mdoc_node *n;
 708         const char       *cp;
 709         enum mdoct        tok;
 710         int               i, sv;
 711         char              mac[5];
 712
 713         sv = offs;
 714
 715         /*
 716          * Copy the first word into a nil-terminated buffer.
 717          * Stop when a space, tab, escape, or eoln is encountered.
 718          */
 719
 720         i = 0;
 721         while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
 722                 mac[i++] = buf[offs++];
 723
 724         mac[i] = '\0';
 725
 726         tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 727
 728         if (tok == MDOC_MAX) {
 729                 mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
 730                     ln, sv, buf + sv - 1);
 731                 return(1);
 732         }
 733
 734         /* Skip a leading escape sequence or tab. */
 735
 736         switch (buf[offs]) {
 737         case '\\':
 738                 cp = buf + offs + 1;
 739                 mandoc_escape(&cp, NULL, NULL);
 740                 offs = cp - buf;
 741                 break;
 742         case '\t':
 743                 offs++;
 744                 break;
 745         default:
 746                 break;
 747         }
 748
 749         /* Jump to the next non-whitespace word. */
 750
 751         while (buf[offs] && ' ' == buf[offs])
 752                 offs++;
 753
 754         /*
 755          * Trailing whitespace.  Note that tabs are allowed to be passed
 756          * into the parser as "text", so we only warn about spaces here.
 757          */
 758
 759         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 760                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 761                     ln, offs - 1, NULL);
 762
 763         /*
 764          * If an initial macro or a list invocation, divert directly
 765          * into macro processing.
 766          */
 767
 768         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok) {
 769                 mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
 770                 return(1);
 771         }
 772
 773         n = mdoc->last;
 774         assert(mdoc->last);
 775
 776         /*
 777          * If the first macro of a `Bl -column', open an `It' block
 778          * context around the parsed macro.
 779          */
 780
 781         if (n->tok == MDOC_Bl && n->type == MDOC_BODY &&
 782             n->end == ENDBODY_NOT && n->norm->Bl.type == LIST_column) {
 783                 mdoc->flags |= MDOC_FREECOL;
 784                 mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
 785                 return(1);
 786         }
 787
 788         /*
 789          * If we're following a block-level `It' within a `Bl -column'
 790          * context (perhaps opened in the above block or in ptext()),
 791          * then open an `It' block context around the parsed macro.
 792          */
 793
 794         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 795             NULL != n->parent &&
 796             MDOC_Bl == n->parent->tok &&
 797             LIST_column == n->parent->norm->Bl.type) {
 798                 mdoc->flags |= MDOC_FREECOL;
 799                 mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf);
 800                 return(1);
 801         }
 802
 803         /* Normal processing of a macro. */
 804
 805         mdoc_macro(mdoc, tok, ln, sv, &offs, buf);
 806
 807         /* In quick mode (for mandocdb), abort after the NAME section. */
 808
 809         if (mdoc->quick && MDOC_Sh == tok &&
 810             SEC_NAME != mdoc->last->sec)
 811                 return(2);
 812
 813         return(1);
 814 }
 815
 816 enum mdelim
 817 mdoc_isdelim(const char *p)
 818 {
 819
 820         if ('\0' == p[0])
 821                 return(DELIM_NONE);
 822
 823         if ('\0' == p[1])
 824                 switch (p[0]) {
 825                 case '(':
 826                         /* FALLTHROUGH */
 827                 case '[':
 828                         return(DELIM_OPEN);
 829                 case '|':
 830                         return(DELIM_MIDDLE);
 831                 case '.':
 832                         /* FALLTHROUGH */
 833                 case ',':
 834                         /* FALLTHROUGH */
 835                 case ';':
 836                         /* FALLTHROUGH */
 837                 case ':':
 838                         /* FALLTHROUGH */
 839                 case '?':
 840                         /* FALLTHROUGH */
 841                 case '!':
 842                         /* FALLTHROUGH */
 843                 case ')':
 844                         /* FALLTHROUGH */
 845                 case ']':
 846                         return(DELIM_CLOSE);
 847                 default:
 848                         return(DELIM_NONE);
 849                 }
 850
 851         if ('\\' != p[0])
 852                 return(DELIM_NONE);
 853
 854         if (0 == strcmp(p + 1, "."))
 855                 return(DELIM_CLOSE);
 856         if (0 == strcmp(p + 1, "fR|\\fP"))
 857                 return(DELIM_MIDDLE);
 858
 859         return(DELIM_NONE);
 860 }
 861
 862 void
 863 mdoc_deroff(char **dest, const struct mdoc_node *n)
 864 {
 865         char    *cp;
 866         size_t   sz;
 867
 868         if (MDOC_TEXT != n->type) {
 869                 for (n = n->child; n; n = n->next)
 870                         mdoc_deroff(dest, n);
 871                 return;
 872         }
 873
 874         /* Skip leading whitespace. */
 875
 876         for (cp = n->string; '\0' != *cp; cp++)
 877                 if (0 == isspace((unsigned char)*cp))
 878                         break;
 879
 880         /* Skip trailing whitespace. */
 881
 882         for (sz = strlen(cp); sz; sz--)
 883                 if (0 == isspace((unsigned char)cp[sz-1]))
 884                         break;
 885
 886         /* Skip empty strings. */
 887
 888         if (0 == sz)
 889                 return;
 890
 891         if (NULL == *dest) {
 892                 *dest = mandoc_strndup(cp, sz);
 893                 return;
 894         }
 895
 896         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 897         free(*dest);
 898         *dest = cp;
 899 }