mdoc.c

   1 /*      $Id: mdoc.c,v 1.230 2014/11/28 01:05:43 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  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 int
 194 mdoc_endparse(struct mdoc *mdoc)
 195 {
 196
 197         return(mdoc_macroend(mdoc));
 198 }
 199
 200 int
 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         return(1);
 212 }
 213
 214 int
 215 mdoc_addspan(struct mdoc *mdoc, const struct tbl_span *sp)
 216 {
 217         struct mdoc_node *n;
 218
 219         n = node_alloc(mdoc, sp->line, 0, MDOC_MAX, MDOC_TBL);
 220         n->span = sp;
 221         node_append(mdoc, n);
 222         mdoc->next = MDOC_NEXT_SIBLING;
 223         return(1);
 224 }
 225
 226 /*
 227  * Main parse routine.  Parses a single line -- really just hands off to
 228  * the macro (mdoc_pmacro()) or text parser (mdoc_ptext()).
 229  */
 230 int
 231 mdoc_parseln(struct mdoc *mdoc, int ln, char *buf, int offs)
 232 {
 233
 234         if (mdoc->last->type != MDOC_EQN || ln > mdoc->last->line)
 235                 mdoc->flags |= MDOC_NEWLINE;
 236
 237         /*
 238          * Let the roff nS register switch SYNOPSIS mode early,
 239          * such that the parser knows at all times
 240          * whether this mode is on or off.
 241          * Note that this mode is also switched by the Sh macro.
 242          */
 243         if (roff_getreg(mdoc->roff, "nS"))
 244                 mdoc->flags |= MDOC_SYNOPSIS;
 245         else
 246                 mdoc->flags &= ~MDOC_SYNOPSIS;
 247
 248         return(roff_getcontrol(mdoc->roff, buf, &offs) ?
 249             mdoc_pmacro(mdoc, ln, buf, offs) :
 250             mdoc_ptext(mdoc, ln, buf, offs));
 251 }
 252
 253 int
 254 mdoc_macro(MACRO_PROT_ARGS)
 255 {
 256         assert(tok < MDOC_MAX);
 257
 258         if (mdoc->flags & MDOC_PBODY) {
 259                 if (tok == MDOC_Dt) {
 260                         mandoc_vmsg(MANDOCERR_DT_LATE,
 261                             mdoc->parse, line, ppos,
 262                             "Dt %s", buf + *pos);
 263                         return(1);
 264                 }
 265         } else if ( ! (mdoc_macros[tok].flags & MDOC_PROLOGUE)) {
 266                 if (mdoc->meta.title == NULL) {
 267                         mandoc_vmsg(MANDOCERR_DT_NOTITLE,
 268                             mdoc->parse, line, ppos, "%s %s",
 269                             mdoc_macronames[tok], buf + *pos);
 270                         mdoc->meta.title = mandoc_strdup("UNTITLED");
 271                 }
 272                 if (NULL == mdoc->meta.vol)
 273                         mdoc->meta.vol = mandoc_strdup("LOCAL");
 274                 mdoc->flags |= MDOC_PBODY;
 275         }
 276
 277         return((*mdoc_macros[tok].fp)(mdoc, tok, line, ppos, pos, buf));
 278 }
 279
 280
 281 static void
 282 node_append(struct mdoc *mdoc, struct mdoc_node *p)
 283 {
 284
 285         assert(mdoc->last);
 286         assert(mdoc->first);
 287         assert(MDOC_ROOT != p->type);
 288
 289         switch (mdoc->next) {
 290         case MDOC_NEXT_SIBLING:
 291                 mdoc->last->next = p;
 292                 p->prev = mdoc->last;
 293                 p->parent = mdoc->last->parent;
 294                 break;
 295         case MDOC_NEXT_CHILD:
 296                 mdoc->last->child = p;
 297                 p->parent = mdoc->last;
 298                 break;
 299         default:
 300                 abort();
 301                 /* NOTREACHED */
 302         }
 303
 304         p->parent->nchild++;
 305
 306         /*
 307          * Copy over the normalised-data pointer of our parent.  Not
 308          * everybody has one, but copying a null pointer is fine.
 309          */
 310
 311         switch (p->type) {
 312         case MDOC_BODY:
 313                 if (ENDBODY_NOT != p->end)
 314                         break;
 315                 /* FALLTHROUGH */
 316         case MDOC_TAIL:
 317                 /* FALLTHROUGH */
 318         case MDOC_HEAD:
 319                 p->norm = p->parent->norm;
 320                 break;
 321         default:
 322                 break;
 323         }
 324
 325         mdoc_valid_pre(mdoc, p);
 326
 327         switch (p->type) {
 328         case MDOC_HEAD:
 329                 assert(MDOC_BLOCK == p->parent->type);
 330                 p->parent->head = p;
 331                 break;
 332         case MDOC_TAIL:
 333                 assert(MDOC_BLOCK == p->parent->type);
 334                 p->parent->tail = p;
 335                 break;
 336         case MDOC_BODY:
 337                 if (p->end)
 338                         break;
 339                 assert(MDOC_BLOCK == p->parent->type);
 340                 p->parent->body = p;
 341                 break;
 342         default:
 343                 break;
 344         }
 345
 346         mdoc->last = p;
 347
 348         switch (p->type) {
 349         case MDOC_TBL:
 350                 /* FALLTHROUGH */
 351         case MDOC_TEXT:
 352                 mdoc_valid_post(mdoc);
 353                 break;
 354         default:
 355                 break;
 356         }
 357 }
 358
 359 static struct mdoc_node *
 360 node_alloc(struct mdoc *mdoc, int line, int pos,
 361                 enum mdoct tok, enum mdoc_type type)
 362 {
 363         struct mdoc_node *p;
 364
 365         p = mandoc_calloc(1, sizeof(struct mdoc_node));
 366         p->sec = mdoc->lastsec;
 367         p->line = line;
 368         p->pos = pos;
 369         p->lastline = line;
 370         p->tok = tok;
 371         p->type = type;
 372
 373         /* Flag analysis. */
 374
 375         if (MDOC_SYNOPSIS & mdoc->flags)
 376                 p->flags |= MDOC_SYNPRETTY;
 377         else
 378                 p->flags &= ~MDOC_SYNPRETTY;
 379         if (MDOC_NEWLINE & mdoc->flags)
 380                 p->flags |= MDOC_LINE;
 381         mdoc->flags &= ~MDOC_NEWLINE;
 382
 383         return(p);
 384 }
 385
 386 int
 387 mdoc_tail_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 388 {
 389         struct mdoc_node *p;
 390
 391         p = node_alloc(mdoc, line, pos, tok, MDOC_TAIL);
 392         node_append(mdoc, p);
 393         mdoc->next = MDOC_NEXT_CHILD;
 394         return(1);
 395 }
 396
 397 int
 398 mdoc_head_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 399 {
 400         struct mdoc_node *p;
 401
 402         assert(mdoc->first);
 403         assert(mdoc->last);
 404
 405         p = node_alloc(mdoc, line, pos, tok, MDOC_HEAD);
 406         node_append(mdoc, p);
 407         mdoc->next = MDOC_NEXT_CHILD;
 408         return(1);
 409 }
 410
 411 int
 412 mdoc_body_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok)
 413 {
 414         struct mdoc_node *p;
 415
 416         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 417         node_append(mdoc, p);
 418         mdoc->next = MDOC_NEXT_CHILD;
 419         return(1);
 420 }
 421
 422 int
 423 mdoc_endbody_alloc(struct mdoc *mdoc, int line, int pos, enum mdoct tok,
 424                 struct mdoc_node *body, enum mdoc_endbody end)
 425 {
 426         struct mdoc_node *p;
 427
 428         p = node_alloc(mdoc, line, pos, tok, MDOC_BODY);
 429         p->pending = body;
 430         p->norm = body->norm;
 431         p->end = end;
 432         node_append(mdoc, p);
 433         mdoc->next = MDOC_NEXT_SIBLING;
 434         return(1);
 435 }
 436
 437 int
 438 mdoc_block_alloc(struct mdoc *mdoc, int line, int pos,
 439                 enum mdoct tok, struct mdoc_arg *args)
 440 {
 441         struct mdoc_node *p;
 442
 443         p = node_alloc(mdoc, line, pos, tok, MDOC_BLOCK);
 444         p->args = args;
 445         if (p->args)
 446                 (args->refcnt)++;
 447
 448         switch (tok) {
 449         case MDOC_Bd:
 450                 /* FALLTHROUGH */
 451         case MDOC_Bf:
 452                 /* FALLTHROUGH */
 453         case MDOC_Bl:
 454                 /* FALLTHROUGH */
 455         case MDOC_En:
 456                 /* FALLTHROUGH */
 457         case MDOC_Rs:
 458                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 459                 break;
 460         default:
 461                 break;
 462         }
 463         node_append(mdoc, p);
 464         mdoc->next = MDOC_NEXT_CHILD;
 465         return(1);
 466 }
 467
 468 int
 469 mdoc_elem_alloc(struct mdoc *mdoc, int line, int pos,
 470                 enum mdoct tok, struct mdoc_arg *args)
 471 {
 472         struct mdoc_node *p;
 473
 474         p = node_alloc(mdoc, line, pos, tok, MDOC_ELEM);
 475         p->args = args;
 476         if (p->args)
 477                 (args->refcnt)++;
 478
 479         switch (tok) {
 480         case MDOC_An:
 481                 p->norm = mandoc_calloc(1, sizeof(union mdoc_data));
 482                 break;
 483         default:
 484                 break;
 485         }
 486         node_append(mdoc, p);
 487         mdoc->next = MDOC_NEXT_CHILD;
 488         return(1);
 489 }
 490
 491 int
 492 mdoc_word_alloc(struct mdoc *mdoc, int line, int pos, const char *p)
 493 {
 494         struct mdoc_node *n;
 495
 496         n = node_alloc(mdoc, line, pos, MDOC_MAX, MDOC_TEXT);
 497         n->string = roff_strdup(mdoc->roff, p);
 498         node_append(mdoc, n);
 499         mdoc->next = MDOC_NEXT_SIBLING;
 500         return(1);
 501 }
 502
 503 void
 504 mdoc_word_append(struct mdoc *mdoc, const char *p)
 505 {
 506         struct mdoc_node        *n;
 507         char                    *addstr, *newstr;
 508
 509         n = mdoc->last;
 510         addstr = roff_strdup(mdoc->roff, p);
 511         mandoc_asprintf(&newstr, "%s %s", n->string, addstr);
 512         free(addstr);
 513         free(n->string);
 514         n->string = newstr;
 515         mdoc->next = MDOC_NEXT_SIBLING;
 516 }
 517
 518 static void
 519 mdoc_node_free(struct mdoc_node *p)
 520 {
 521
 522         if (MDOC_BLOCK == p->type || MDOC_ELEM == p->type)
 523                 free(p->norm);
 524         if (p->string)
 525                 free(p->string);
 526         if (p->args)
 527                 mdoc_argv_free(p->args);
 528         free(p);
 529 }
 530
 531 static void
 532 mdoc_node_unlink(struct mdoc *mdoc, struct mdoc_node *n)
 533 {
 534
 535         /* Adjust siblings. */
 536
 537         if (n->prev)
 538                 n->prev->next = n->next;
 539         if (n->next)
 540                 n->next->prev = n->prev;
 541
 542         /* Adjust parent. */
 543
 544         if (n->parent) {
 545                 n->parent->nchild--;
 546                 if (n->parent->child == n)
 547                         n->parent->child = n->prev ? n->prev : n->next;
 548                 if (n->parent->last == n)
 549                         n->parent->last = n->prev ? n->prev : NULL;
 550         }
 551
 552         /* Adjust parse point, if applicable. */
 553
 554         if (mdoc && mdoc->last == n) {
 555                 if (n->prev) {
 556                         mdoc->last = n->prev;
 557                         mdoc->next = MDOC_NEXT_SIBLING;
 558                 } else {
 559                         mdoc->last = n->parent;
 560                         mdoc->next = MDOC_NEXT_CHILD;
 561                 }
 562         }
 563
 564         if (mdoc && mdoc->first == n)
 565                 mdoc->first = NULL;
 566 }
 567
 568 void
 569 mdoc_node_delete(struct mdoc *mdoc, struct mdoc_node *p)
 570 {
 571
 572         while (p->child) {
 573                 assert(p->nchild);
 574                 mdoc_node_delete(mdoc, p->child);
 575         }
 576         assert(0 == p->nchild);
 577
 578         mdoc_node_unlink(mdoc, p);
 579         mdoc_node_free(p);
 580 }
 581
 582 int
 583 mdoc_node_relink(struct mdoc *mdoc, struct mdoc_node *p)
 584 {
 585
 586         mdoc_node_unlink(mdoc, p);
 587         node_append(mdoc, p);
 588         return(1);
 589 }
 590
 591 /*
 592  * Parse free-form text, that is, a line that does not begin with the
 593  * control character.
 594  */
 595 static int
 596 mdoc_ptext(struct mdoc *mdoc, int line, char *buf, int offs)
 597 {
 598         char             *c, *ws, *end;
 599         struct mdoc_node *n;
 600
 601         assert(mdoc->last);
 602         n = mdoc->last;
 603
 604         /*
 605          * Divert directly to list processing if we're encountering a
 606          * columnar MDOC_BLOCK with or without a prior MDOC_BLOCK entry
 607          * (a MDOC_BODY means it's already open, in which case we should
 608          * process within its context in the normal way).
 609          */
 610
 611         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 612             LIST_column == n->norm->Bl.type) {
 613                 /* `Bl' is open without any children. */
 614                 mdoc->flags |= MDOC_FREECOL;
 615                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 616         }
 617
 618         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 619             NULL != n->parent &&
 620             MDOC_Bl == n->parent->tok &&
 621             LIST_column == n->parent->norm->Bl.type) {
 622                 /* `Bl' has block-level `It' children. */
 623                 mdoc->flags |= MDOC_FREECOL;
 624                 return(mdoc_macro(mdoc, MDOC_It, line, offs, &offs, buf));
 625         }
 626
 627         /*
 628          * Search for the beginning of unescaped trailing whitespace (ws)
 629          * and for the first character not to be output (end).
 630          */
 631
 632         /* FIXME: replace with strcspn(). */
 633         ws = NULL;
 634         for (c = end = buf + offs; *c; c++) {
 635                 switch (*c) {
 636                 case ' ':
 637                         if (NULL == ws)
 638                                 ws = c;
 639                         continue;
 640                 case '\t':
 641                         /*
 642                          * Always warn about trailing tabs,
 643                          * even outside literal context,
 644                          * where they should be put on the next line.
 645                          */
 646                         if (NULL == ws)
 647                                 ws = c;
 648                         /*
 649                          * Strip trailing tabs in literal context only;
 650                          * outside, they affect the next line.
 651                          */
 652                         if (MDOC_LITERAL & mdoc->flags)
 653                                 continue;
 654                         break;
 655                 case '\\':
 656                         /* Skip the escaped character, too, if any. */
 657                         if (c[1])
 658                                 c++;
 659                         /* FALLTHROUGH */
 660                 default:
 661                         ws = NULL;
 662                         break;
 663                 }
 664                 end = c + 1;
 665         }
 666         *end = '\0';
 667
 668         if (ws)
 669                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 670                     line, (int)(ws-buf), NULL);
 671
 672         if ('\0' == buf[offs] && ! (MDOC_LITERAL & mdoc->flags)) {
 673                 mandoc_msg(MANDOCERR_FI_BLANK, mdoc->parse,
 674                     line, (int)(c - buf), NULL);
 675
 676                 /*
 677                  * Insert a `sp' in the case of a blank line.  Technically,
 678                  * blank lines aren't allowed, but enough manuals assume this
 679                  * behaviour that we want to work around it.
 680                  */
 681                 if ( ! mdoc_elem_alloc(mdoc, line, offs, MDOC_sp, NULL))
 682                         return(0);
 683
 684                 mdoc->next = MDOC_NEXT_SIBLING;
 685
 686                 mdoc_valid_post(mdoc);
 687                 return(1);
 688         }
 689
 690         if ( ! mdoc_word_alloc(mdoc, line, offs, buf+offs))
 691                 return(0);
 692
 693         if (MDOC_LITERAL & mdoc->flags)
 694                 return(1);
 695
 696         /*
 697          * End-of-sentence check.  If the last character is an unescaped
 698          * EOS character, then flag the node as being the end of a
 699          * sentence.  The front-end will know how to interpret this.
 700          */
 701
 702         assert(buf < end);
 703
 704         if (mandoc_eos(buf+offs, (size_t)(end-buf-offs)))
 705                 mdoc->last->flags |= MDOC_EOS;
 706
 707         return(1);
 708 }
 709
 710 /*
 711  * Parse a macro line, that is, a line beginning with the control
 712  * character.
 713  */
 714 static int
 715 mdoc_pmacro(struct mdoc *mdoc, int ln, char *buf, int offs)
 716 {
 717         struct mdoc_node *n;
 718         const char       *cp;
 719         enum mdoct        tok;
 720         int               i, sv;
 721         char              mac[5];
 722
 723         sv = offs;
 724
 725         /*
 726          * Copy the first word into a nil-terminated buffer.
 727          * Stop when a space, tab, escape, or eoln is encountered.
 728          */
 729
 730         i = 0;
 731         while (i < 4 && strchr(" \t\\", buf[offs]) == NULL)
 732                 mac[i++] = buf[offs++];
 733
 734         mac[i] = '\0';
 735
 736         tok = (i > 1 && i < 4) ? mdoc_hash_find(mac) : MDOC_MAX;
 737
 738         if (tok == MDOC_MAX) {
 739                 mandoc_msg(MANDOCERR_MACRO, mdoc->parse,
 740                     ln, sv, buf + sv - 1);
 741                 return(1);
 742         }
 743
 744         /* Skip a leading escape sequence or tab. */
 745
 746         switch (buf[offs]) {
 747         case '\\':
 748                 cp = buf + offs + 1;
 749                 mandoc_escape(&cp, NULL, NULL);
 750                 offs = cp - buf;
 751                 break;
 752         case '\t':
 753                 offs++;
 754                 break;
 755         default:
 756                 break;
 757         }
 758
 759         /* Jump to the next non-whitespace word. */
 760
 761         while (buf[offs] && ' ' == buf[offs])
 762                 offs++;
 763
 764         /*
 765          * Trailing whitespace.  Note that tabs are allowed to be passed
 766          * into the parser as "text", so we only warn about spaces here.
 767          */
 768
 769         if ('\0' == buf[offs] && ' ' == buf[offs - 1])
 770                 mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 771                     ln, offs - 1, NULL);
 772
 773         /*
 774          * If an initial macro or a list invocation, divert directly
 775          * into macro processing.
 776          */
 777
 778         if (NULL == mdoc->last || MDOC_It == tok || MDOC_El == tok)
 779                 return(mdoc_macro(mdoc, tok, ln, sv, &offs, buf));
 780
 781         n = mdoc->last;
 782         assert(mdoc->last);
 783
 784         /*
 785          * If the first macro of a `Bl -column', open an `It' block
 786          * context around the parsed macro.
 787          */
 788
 789         if (MDOC_Bl == n->tok && MDOC_BODY == n->type &&
 790             LIST_column == n->norm->Bl.type) {
 791                 mdoc->flags |= MDOC_FREECOL;
 792                 return(mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf));
 793         }
 794
 795         /*
 796          * If we're following a block-level `It' within a `Bl -column'
 797          * context (perhaps opened in the above block or in ptext()),
 798          * then open an `It' block context around the parsed macro.
 799          */
 800
 801         if (MDOC_It == n->tok && MDOC_BLOCK == n->type &&
 802             NULL != n->parent &&
 803             MDOC_Bl == n->parent->tok &&
 804             LIST_column == n->parent->norm->Bl.type) {
 805                 mdoc->flags |= MDOC_FREECOL;
 806                 return(mdoc_macro(mdoc, MDOC_It, ln, sv, &sv, buf));
 807         }
 808
 809         /* Normal processing of a macro. */
 810
 811         if ( ! mdoc_macro(mdoc, tok, ln, sv, &offs, buf))
 812                 return(0);
 813
 814         /* In quick mode (for mandocdb), abort after the NAME section. */
 815
 816         if (mdoc->quick && MDOC_Sh == tok &&
 817             SEC_NAME != mdoc->last->sec)
 818                 return(2);
 819
 820         return(1);
 821 }
 822
 823 enum mdelim
 824 mdoc_isdelim(const char *p)
 825 {
 826
 827         if ('\0' == p[0])
 828                 return(DELIM_NONE);
 829
 830         if ('\0' == p[1])
 831                 switch (p[0]) {
 832                 case '(':
 833                         /* FALLTHROUGH */
 834                 case '[':
 835                         return(DELIM_OPEN);
 836                 case '|':
 837                         return(DELIM_MIDDLE);
 838                 case '.':
 839                         /* FALLTHROUGH */
 840                 case ',':
 841                         /* FALLTHROUGH */
 842                 case ';':
 843                         /* FALLTHROUGH */
 844                 case ':':
 845                         /* FALLTHROUGH */
 846                 case '?':
 847                         /* FALLTHROUGH */
 848                 case '!':
 849                         /* FALLTHROUGH */
 850                 case ')':
 851                         /* FALLTHROUGH */
 852                 case ']':
 853                         return(DELIM_CLOSE);
 854                 default:
 855                         return(DELIM_NONE);
 856                 }
 857
 858         if ('\\' != p[0])
 859                 return(DELIM_NONE);
 860
 861         if (0 == strcmp(p + 1, "."))
 862                 return(DELIM_CLOSE);
 863         if (0 == strcmp(p + 1, "fR|\\fP"))
 864                 return(DELIM_MIDDLE);
 865
 866         return(DELIM_NONE);
 867 }
 868
 869 void
 870 mdoc_deroff(char **dest, const struct mdoc_node *n)
 871 {
 872         char    *cp;
 873         size_t   sz;
 874
 875         if (MDOC_TEXT != n->type) {
 876                 for (n = n->child; n; n = n->next)
 877                         mdoc_deroff(dest, n);
 878                 return;
 879         }
 880
 881         /* Skip leading whitespace. */
 882
 883         for (cp = n->string; '\0' != *cp; cp++)
 884                 if (0 == isspace((unsigned char)*cp))
 885                         break;
 886
 887         /* Skip trailing whitespace. */
 888
 889         for (sz = strlen(cp); sz; sz--)
 890                 if (0 == isspace((unsigned char)cp[sz-1]))
 891                         break;
 892
 893         /* Skip empty strings. */
 894
 895         if (0 == sz)
 896                 return;
 897
 898         if (NULL == *dest) {
 899                 *dest = mandoc_strndup(cp, sz);
 900                 return;
 901         }
 902
 903         mandoc_asprintf(&cp, "%s %*s", *dest, (int)sz, cp);
 904         free(*dest);
 905         *dest = cp;
 906 }