mdoc_argv.c

   1 /*      $Id: mdoc_argv.c,v 1.96 2014/08/10 23:54:41 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
   4  * Copyright (c) 2012 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 <stdlib.h>
  24 #include <stdio.h>
  25 #include <string.h>
  26
  27 #include "mdoc.h"
  28 #include "mandoc.h"
  29 #include "mandoc_aux.h"
  30 #include "libmdoc.h"
  31 #include "libmandoc.h"
  32
  33 #define MULTI_STEP       5 /* pre-allocate argument values */
  34 #define DELIMSZ          6 /* max possible size of a delimiter */
  35
  36 enum    argsflag {
  37         ARGSFL_NONE = 0,
  38         ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
  39         ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
  40 };
  41
  42 enum    argvflag {
  43         ARGV_NONE, /* no args to flag (e.g., -split) */
  44         ARGV_SINGLE, /* one arg to flag (e.g., -file xxx)  */
  45         ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
  46 };
  47
  48 struct  mdocarg {
  49         enum argsflag    flags;
  50         const enum mdocargt *argvs;
  51 };
  52
  53 static  void             argn_free(struct mdoc_arg *, int);
  54 static  enum margserr    args(struct mdoc *, int, int *,
  55                                 char *, enum argsflag, char **);
  56 static  int              args_checkpunct(const char *, int);
  57 static  int              argv_multi(struct mdoc *, int,
  58                                 struct mdoc_argv *, int *, char *);
  59 static  int              argv_single(struct mdoc *, int,
  60                                 struct mdoc_argv *, int *, char *);
  61
  62 static  const enum argvflag argvflags[MDOC_ARG_MAX] = {
  63         ARGV_NONE,      /* MDOC_Split */
  64         ARGV_NONE,      /* MDOC_Nosplit */
  65         ARGV_NONE,      /* MDOC_Ragged */
  66         ARGV_NONE,      /* MDOC_Unfilled */
  67         ARGV_NONE,      /* MDOC_Literal */
  68         ARGV_SINGLE,    /* MDOC_File */
  69         ARGV_SINGLE,    /* MDOC_Offset */
  70         ARGV_NONE,      /* MDOC_Bullet */
  71         ARGV_NONE,      /* MDOC_Dash */
  72         ARGV_NONE,      /* MDOC_Hyphen */
  73         ARGV_NONE,      /* MDOC_Item */
  74         ARGV_NONE,      /* MDOC_Enum */
  75         ARGV_NONE,      /* MDOC_Tag */
  76         ARGV_NONE,      /* MDOC_Diag */
  77         ARGV_NONE,      /* MDOC_Hang */
  78         ARGV_NONE,      /* MDOC_Ohang */
  79         ARGV_NONE,      /* MDOC_Inset */
  80         ARGV_MULTI,     /* MDOC_Column */
  81         ARGV_SINGLE,    /* MDOC_Width */
  82         ARGV_NONE,      /* MDOC_Compact */
  83         ARGV_NONE,      /* MDOC_Std */
  84         ARGV_NONE,      /* MDOC_Filled */
  85         ARGV_NONE,      /* MDOC_Words */
  86         ARGV_NONE,      /* MDOC_Emphasis */
  87         ARGV_NONE,      /* MDOC_Symbolic */
  88         ARGV_NONE       /* MDOC_Symbolic */
  89 };
  90
  91 static  const enum mdocargt args_Ex[] = {
  92         MDOC_Std,
  93         MDOC_ARG_MAX
  94 };
  95
  96 static  const enum mdocargt args_An[] = {
  97         MDOC_Split,
  98         MDOC_Nosplit,
  99         MDOC_ARG_MAX
 100 };
 101
 102 static  const enum mdocargt args_Bd[] = {
 103         MDOC_Ragged,
 104         MDOC_Unfilled,
 105         MDOC_Filled,
 106         MDOC_Literal,
 107         MDOC_File,
 108         MDOC_Offset,
 109         MDOC_Compact,
 110         MDOC_Centred,
 111         MDOC_ARG_MAX
 112 };
 113
 114 static  const enum mdocargt args_Bf[] = {
 115         MDOC_Emphasis,
 116         MDOC_Literal,
 117         MDOC_Symbolic,
 118         MDOC_ARG_MAX
 119 };
 120
 121 static  const enum mdocargt args_Bk[] = {
 122         MDOC_Words,
 123         MDOC_ARG_MAX
 124 };
 125
 126 static  const enum mdocargt args_Bl[] = {
 127         MDOC_Bullet,
 128         MDOC_Dash,
 129         MDOC_Hyphen,
 130         MDOC_Item,
 131         MDOC_Enum,
 132         MDOC_Tag,
 133         MDOC_Diag,
 134         MDOC_Hang,
 135         MDOC_Ohang,
 136         MDOC_Inset,
 137         MDOC_Column,
 138         MDOC_Width,
 139         MDOC_Offset,
 140         MDOC_Compact,
 141         MDOC_Nested,
 142         MDOC_ARG_MAX
 143 };
 144
 145 static  const struct mdocarg mdocargs[MDOC_MAX] = {
 146         { ARGSFL_DELIM, NULL }, /* Ap */
 147         { ARGSFL_NONE, NULL }, /* Dd */
 148         { ARGSFL_NONE, NULL }, /* Dt */
 149         { ARGSFL_NONE, NULL }, /* Os */
 150         { ARGSFL_NONE, NULL }, /* Sh */
 151         { ARGSFL_NONE, NULL }, /* Ss */
 152         { ARGSFL_NONE, NULL }, /* Pp */
 153         { ARGSFL_DELIM, NULL }, /* D1 */
 154         { ARGSFL_DELIM, NULL }, /* Dl */
 155         { ARGSFL_NONE, args_Bd }, /* Bd */
 156         { ARGSFL_NONE, NULL }, /* Ed */
 157         { ARGSFL_NONE, args_Bl }, /* Bl */
 158         { ARGSFL_NONE, NULL }, /* El */
 159         { ARGSFL_NONE, NULL }, /* It */
 160         { ARGSFL_DELIM, NULL }, /* Ad */
 161         { ARGSFL_DELIM, args_An }, /* An */
 162         { ARGSFL_DELIM, NULL }, /* Ar */
 163         { ARGSFL_DELIM, NULL }, /* Cd */
 164         { ARGSFL_DELIM, NULL }, /* Cm */
 165         { ARGSFL_DELIM, NULL }, /* Dv */
 166         { ARGSFL_DELIM, NULL }, /* Er */
 167         { ARGSFL_DELIM, NULL }, /* Ev */
 168         { ARGSFL_NONE, args_Ex }, /* Ex */
 169         { ARGSFL_DELIM, NULL }, /* Fa */
 170         { ARGSFL_NONE, NULL }, /* Fd */
 171         { ARGSFL_DELIM, NULL }, /* Fl */
 172         { ARGSFL_DELIM, NULL }, /* Fn */
 173         { ARGSFL_DELIM, NULL }, /* Ft */
 174         { ARGSFL_DELIM, NULL }, /* Ic */
 175         { ARGSFL_DELIM, NULL }, /* In */
 176         { ARGSFL_DELIM, NULL }, /* Li */
 177         { ARGSFL_NONE, NULL }, /* Nd */
 178         { ARGSFL_DELIM, NULL }, /* Nm */
 179         { ARGSFL_DELIM, NULL }, /* Op */
 180         { ARGSFL_DELIM, NULL }, /* Ot */
 181         { ARGSFL_DELIM, NULL }, /* Pa */
 182         { ARGSFL_NONE, args_Ex }, /* Rv */
 183         { ARGSFL_DELIM, NULL }, /* St */
 184         { ARGSFL_DELIM, NULL }, /* Va */
 185         { ARGSFL_DELIM, NULL }, /* Vt */
 186         { ARGSFL_DELIM, NULL }, /* Xr */
 187         { ARGSFL_NONE, NULL }, /* %A */
 188         { ARGSFL_NONE, NULL }, /* %B */
 189         { ARGSFL_NONE, NULL }, /* %D */
 190         { ARGSFL_NONE, NULL }, /* %I */
 191         { ARGSFL_NONE, NULL }, /* %J */
 192         { ARGSFL_NONE, NULL }, /* %N */
 193         { ARGSFL_NONE, NULL }, /* %O */
 194         { ARGSFL_NONE, NULL }, /* %P */
 195         { ARGSFL_NONE, NULL }, /* %R */
 196         { ARGSFL_NONE, NULL }, /* %T */
 197         { ARGSFL_NONE, NULL }, /* %V */
 198         { ARGSFL_DELIM, NULL }, /* Ac */
 199         { ARGSFL_NONE, NULL }, /* Ao */
 200         { ARGSFL_DELIM, NULL }, /* Aq */
 201         { ARGSFL_DELIM, NULL }, /* At */
 202         { ARGSFL_DELIM, NULL }, /* Bc */
 203         { ARGSFL_NONE, args_Bf }, /* Bf */
 204         { ARGSFL_NONE, NULL }, /* Bo */
 205         { ARGSFL_DELIM, NULL }, /* Bq */
 206         { ARGSFL_DELIM, NULL }, /* Bsx */
 207         { ARGSFL_DELIM, NULL }, /* Bx */
 208         { ARGSFL_NONE, NULL }, /* Db */
 209         { ARGSFL_DELIM, NULL }, /* Dc */
 210         { ARGSFL_NONE, NULL }, /* Do */
 211         { ARGSFL_DELIM, NULL }, /* Dq */
 212         { ARGSFL_DELIM, NULL }, /* Ec */
 213         { ARGSFL_NONE, NULL }, /* Ef */
 214         { ARGSFL_DELIM, NULL }, /* Em */
 215         { ARGSFL_NONE, NULL }, /* Eo */
 216         { ARGSFL_DELIM, NULL }, /* Fx */
 217         { ARGSFL_DELIM, NULL }, /* Ms */
 218         { ARGSFL_DELIM, NULL }, /* No */
 219         { ARGSFL_DELIM, NULL }, /* Ns */
 220         { ARGSFL_DELIM, NULL }, /* Nx */
 221         { ARGSFL_DELIM, NULL }, /* Ox */
 222         { ARGSFL_DELIM, NULL }, /* Pc */
 223         { ARGSFL_DELIM, NULL }, /* Pf */
 224         { ARGSFL_NONE, NULL }, /* Po */
 225         { ARGSFL_DELIM, NULL }, /* Pq */
 226         { ARGSFL_DELIM, NULL }, /* Qc */
 227         { ARGSFL_DELIM, NULL }, /* Ql */
 228         { ARGSFL_NONE, NULL }, /* Qo */
 229         { ARGSFL_DELIM, NULL }, /* Qq */
 230         { ARGSFL_NONE, NULL }, /* Re */
 231         { ARGSFL_NONE, NULL }, /* Rs */
 232         { ARGSFL_DELIM, NULL }, /* Sc */
 233         { ARGSFL_NONE, NULL }, /* So */
 234         { ARGSFL_DELIM, NULL }, /* Sq */
 235         { ARGSFL_NONE, NULL }, /* Sm */
 236         { ARGSFL_DELIM, NULL }, /* Sx */
 237         { ARGSFL_DELIM, NULL }, /* Sy */
 238         { ARGSFL_DELIM, NULL }, /* Tn */
 239         { ARGSFL_DELIM, NULL }, /* Ux */
 240         { ARGSFL_DELIM, NULL }, /* Xc */
 241         { ARGSFL_NONE, NULL }, /* Xo */
 242         { ARGSFL_NONE, NULL }, /* Fo */
 243         { ARGSFL_DELIM, NULL }, /* Fc */
 244         { ARGSFL_NONE, NULL }, /* Oo */
 245         { ARGSFL_DELIM, NULL }, /* Oc */
 246         { ARGSFL_NONE, args_Bk }, /* Bk */
 247         { ARGSFL_NONE, NULL }, /* Ek */
 248         { ARGSFL_NONE, NULL }, /* Bt */
 249         { ARGSFL_NONE, NULL }, /* Hf */
 250         { ARGSFL_DELIM, NULL }, /* Fr */
 251         { ARGSFL_NONE, NULL }, /* Ud */
 252         { ARGSFL_DELIM, NULL }, /* Lb */
 253         { ARGSFL_NONE, NULL }, /* Lp */
 254         { ARGSFL_DELIM, NULL }, /* Lk */
 255         { ARGSFL_DELIM, NULL }, /* Mt */
 256         { ARGSFL_DELIM, NULL }, /* Brq */
 257         { ARGSFL_NONE, NULL }, /* Bro */
 258         { ARGSFL_DELIM, NULL }, /* Brc */
 259         { ARGSFL_NONE, NULL }, /* %C */
 260         { ARGSFL_NONE, NULL }, /* Es */
 261         { ARGSFL_DELIM, NULL }, /* En */
 262         { ARGSFL_DELIM, NULL }, /* Dx */
 263         { ARGSFL_NONE, NULL }, /* %Q */
 264         { ARGSFL_NONE, NULL }, /* br */
 265         { ARGSFL_NONE, NULL }, /* sp */
 266         { ARGSFL_NONE, NULL }, /* %U */
 267         { ARGSFL_NONE, NULL }, /* Ta */
 268         { ARGSFL_NONE, NULL }, /* ll */
 269 };
 270
 271
 272 /*
 273  * Parse an argument from line text.  This comes in the form of -key
 274  * [value0...], which may either have a single mandatory value, at least
 275  * one mandatory value, an optional single value, or no value.
 276  */
 277 enum margverr
 278 mdoc_argv(struct mdoc *mdoc, int line, enum mdoct tok,
 279                 struct mdoc_arg **v, int *pos, char *buf)
 280 {
 281         char             *p, sv;
 282         struct mdoc_argv tmp;
 283         struct mdoc_arg  *arg;
 284         const enum mdocargt *ap;
 285
 286         if ('\0' == buf[*pos])
 287                 return(ARGV_EOLN);
 288         else if (NULL == (ap = mdocargs[tok].argvs))
 289                 return(ARGV_WORD);
 290         else if ('-' != buf[*pos])
 291                 return(ARGV_WORD);
 292
 293         /* Seek to the first unescaped space. */
 294
 295         p = &buf[++(*pos)];
 296
 297         assert(*pos > 0);
 298
 299         for ( ; buf[*pos] ; (*pos)++)
 300                 if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
 301                         break;
 302
 303         /*
 304          * We want to nil-terminate the word to look it up (it's easier
 305          * that way).  But we may not have a flag, in which case we need
 306          * to restore the line as-is.  So keep around the stray byte,
 307          * which we'll reset upon exiting (if necessary).
 308          */
 309
 310         if ('\0' != (sv = buf[*pos]))
 311                 buf[(*pos)++] = '\0';
 312
 313         /*
 314          * Now look up the word as a flag.  Use temporary storage that
 315          * we'll copy into the node's flags, if necessary.
 316          */
 317
 318         memset(&tmp, 0, sizeof(struct mdoc_argv));
 319
 320         tmp.line = line;
 321         tmp.pos = *pos;
 322         tmp.arg = MDOC_ARG_MAX;
 323
 324         while (MDOC_ARG_MAX != (tmp.arg = *ap++))
 325                 if (0 == strcmp(p, mdoc_argnames[tmp.arg]))
 326                         break;
 327
 328         if (MDOC_ARG_MAX == tmp.arg) {
 329                 /*
 330                  * The flag was not found.
 331                  * Restore saved zeroed byte and return as a word.
 332                  */
 333                 if (sv)
 334                         buf[*pos - 1] = sv;
 335                 return(ARGV_WORD);
 336         }
 337
 338         /* Read to the next word (the argument). */
 339
 340         while (buf[*pos] && ' ' == buf[*pos])
 341                 (*pos)++;
 342
 343         switch (argvflags[tmp.arg]) {
 344         case ARGV_SINGLE:
 345                 if ( ! argv_single(mdoc, line, &tmp, pos, buf))
 346                         return(ARGV_ERROR);
 347                 break;
 348         case ARGV_MULTI:
 349                 if ( ! argv_multi(mdoc, line, &tmp, pos, buf))
 350                         return(ARGV_ERROR);
 351                 break;
 352         case ARGV_NONE:
 353                 break;
 354         }
 355
 356         if (NULL == (arg = *v))
 357                 arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
 358
 359         arg->argc++;
 360         arg->argv = mandoc_reallocarray(arg->argv,
 361             arg->argc, sizeof(struct mdoc_argv));
 362
 363         memcpy(&arg->argv[(int)arg->argc - 1], &tmp,
 364             sizeof(struct mdoc_argv));
 365
 366         return(ARGV_ARG);
 367 }
 368
 369 void
 370 mdoc_argv_free(struct mdoc_arg *p)
 371 {
 372         int              i;
 373
 374         if (NULL == p)
 375                 return;
 376
 377         if (p->refcnt) {
 378                 --(p->refcnt);
 379                 if (p->refcnt)
 380                         return;
 381         }
 382         assert(p->argc);
 383
 384         for (i = (int)p->argc - 1; i >= 0; i--)
 385                 argn_free(p, i);
 386
 387         free(p->argv);
 388         free(p);
 389 }
 390
 391 static void
 392 argn_free(struct mdoc_arg *p, int iarg)
 393 {
 394         struct mdoc_argv *arg;
 395         int               j;
 396
 397         arg = &p->argv[iarg];
 398
 399         if (arg->sz && arg->value) {
 400                 for (j = (int)arg->sz - 1; j >= 0; j--)
 401                         free(arg->value[j]);
 402                 free(arg->value);
 403         }
 404
 405         for (--p->argc; iarg < (int)p->argc; iarg++)
 406                 p->argv[iarg] = p->argv[iarg+1];
 407 }
 408
 409 enum margserr
 410 mdoc_zargs(struct mdoc *mdoc, int line, int *pos, char *buf, char **v)
 411 {
 412
 413         return(args(mdoc, line, pos, buf, ARGSFL_NONE, v));
 414 }
 415
 416 enum margserr
 417 mdoc_args(struct mdoc *mdoc, int line, int *pos,
 418                 char *buf, enum mdoct tok, char **v)
 419 {
 420         enum argsflag     fl;
 421         struct mdoc_node *n;
 422
 423         fl = mdocargs[tok].flags;
 424
 425         if (MDOC_It != tok)
 426                 return(args(mdoc, line, pos, buf, fl, v));
 427
 428         /*
 429          * We know that we're in an `It', so it's reasonable to expect
 430          * us to be sitting in a `Bl'.  Someday this may not be the case
 431          * (if we allow random `It's sitting out there), so provide a
 432          * safe fall-back into the default behaviour.
 433          */
 434
 435         for (n = mdoc->last; n; n = n->parent)
 436                 if (MDOC_Bl == n->tok)
 437                         if (LIST_column == n->norm->Bl.type) {
 438                                 fl = ARGSFL_TABSEP;
 439                                 break;
 440                         }
 441
 442         return(args(mdoc, line, pos, buf, fl, v));
 443 }
 444
 445 static enum margserr
 446 args(struct mdoc *mdoc, int line, int *pos,
 447                 char *buf, enum argsflag fl, char **v)
 448 {
 449         char            *p, *pp;
 450         int              pairs;
 451         enum margserr    rc;
 452
 453         if ('\0' == buf[*pos]) {
 454                 if (MDOC_PPHRASE & mdoc->flags)
 455                         return(ARGS_EOLN);
 456                 /*
 457                  * If we're not in a partial phrase and the flag for
 458                  * being a phrase literal is still set, the punctuation
 459                  * is unterminated.
 460                  */
 461                 if (MDOC_PHRASELIT & mdoc->flags)
 462                         mandoc_msg(MANDOCERR_ARG_QUOTE,
 463                             mdoc->parse, line, *pos, NULL);
 464
 465                 mdoc->flags &= ~MDOC_PHRASELIT;
 466                 return(ARGS_EOLN);
 467         }
 468
 469         *v = &buf[*pos];
 470
 471         if (ARGSFL_DELIM == fl)
 472                 if (args_checkpunct(buf, *pos))
 473                         return(ARGS_PUNCT);
 474
 475         /*
 476          * First handle TABSEP items, restricted to `Bl -column'.  This
 477          * ignores conventional token parsing and instead uses tabs or
 478          * `Ta' macros to separate phrases.  Phrases are parsed again
 479          * for arguments at a later phase.
 480          */
 481
 482         if (ARGSFL_TABSEP == fl) {
 483                 /* Scan ahead to tab (can't be escaped). */
 484                 p = strchr(*v, '\t');
 485                 pp = NULL;
 486
 487                 /* Scan ahead to unescaped `Ta'. */
 488                 if ( ! (MDOC_PHRASELIT & mdoc->flags))
 489                         for (pp = *v; ; pp++) {
 490                                 if (NULL == (pp = strstr(pp, "Ta")))
 491                                         break;
 492                                 if (pp > *v && ' ' != *(pp - 1))
 493                                         continue;
 494                                 if (' ' == *(pp + 2) || '\0' == *(pp + 2))
 495                                         break;
 496                         }
 497
 498                 /* By default, assume a phrase. */
 499                 rc = ARGS_PHRASE;
 500
 501                 /*
 502                  * Adjust new-buffer position to be beyond delimiter
 503                  * mark (e.g., Ta -> end + 2).
 504                  */
 505                 if (p && pp) {
 506                         *pos += pp < p ? 2 : 1;
 507                         rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
 508                         p = pp < p ? pp : p;
 509                 } else if (p && ! pp) {
 510                         rc = ARGS_PPHRASE;
 511                         *pos += 1;
 512                 } else if (pp && ! p) {
 513                         p = pp;
 514                         *pos += 2;
 515                 } else {
 516                         rc = ARGS_PEND;
 517                         p = strchr(*v, 0);
 518                 }
 519
 520                 /* Whitespace check for eoln case... */
 521                 if ('\0' == *p && ' ' == *(p - 1))
 522                         mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 523                             line, *pos, NULL);
 524
 525                 *pos += (int)(p - *v);
 526
 527                 /* Strip delimiter's preceding whitespace. */
 528                 pp = p - 1;
 529                 while (pp > *v && ' ' == *pp) {
 530                         if (pp > *v && '\\' == *(pp - 1))
 531                                 break;
 532                         pp--;
 533                 }
 534                 *(pp + 1) = 0;
 535
 536                 /* Strip delimiter's proceeding whitespace. */
 537                 for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
 538                         /* Skip ahead. */ ;
 539
 540                 return(rc);
 541         }
 542
 543         /*
 544          * Process a quoted literal.  A quote begins with a double-quote
 545          * and ends with a double-quote NOT preceded by a double-quote.
 546          * NUL-terminate the literal in place.
 547          * Collapse pairs of quotes inside quoted literals.
 548          * Whitespace is NOT involved in literal termination.
 549          */
 550
 551         if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
 552                 if ( ! (MDOC_PHRASELIT & mdoc->flags))
 553                         *v = &buf[++(*pos)];
 554
 555                 if (MDOC_PPHRASE & mdoc->flags)
 556                         mdoc->flags |= MDOC_PHRASELIT;
 557
 558                 pairs = 0;
 559                 for ( ; buf[*pos]; (*pos)++) {
 560                         /* Move following text left after quoted quotes. */
 561                         if (pairs)
 562                                 buf[*pos - pairs] = buf[*pos];
 563                         if ('\"' != buf[*pos])
 564                                 continue;
 565                         /* Unquoted quotes end quoted args. */
 566                         if ('\"' != buf[*pos + 1])
 567                                 break;
 568                         /* Quoted quotes collapse. */
 569                         pairs++;
 570                         (*pos)++;
 571                 }
 572                 if (pairs)
 573                         buf[*pos - pairs] = '\0';
 574
 575                 if ('\0' == buf[*pos]) {
 576                         if (MDOC_PPHRASE & mdoc->flags)
 577                                 return(ARGS_QWORD);
 578                         mandoc_msg(MANDOCERR_ARG_QUOTE,
 579                             mdoc->parse, line, *pos, NULL);
 580                         return(ARGS_QWORD);
 581                 }
 582
 583                 mdoc->flags &= ~MDOC_PHRASELIT;
 584                 buf[(*pos)++] = '\0';
 585
 586                 if ('\0' == buf[*pos])
 587                         return(ARGS_QWORD);
 588
 589                 while (' ' == buf[*pos])
 590                         (*pos)++;
 591
 592                 if ('\0' == buf[*pos])
 593                         mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
 594                             line, *pos, NULL);
 595
 596                 return(ARGS_QWORD);
 597         }
 598
 599         p = &buf[*pos];
 600         *v = mandoc_getarg(mdoc->parse, &p, line, pos);
 601
 602         return(ARGS_WORD);
 603 }
 604
 605 /*
 606  * Check if the string consists only of space-separated closing
 607  * delimiters.  This is a bit of a dance: the first must be a close
 608  * delimiter, but it may be followed by middle delimiters.  Arbitrary
 609  * whitespace may separate these tokens.
 610  */
 611 static int
 612 args_checkpunct(const char *buf, int i)
 613 {
 614         int              j;
 615         char             dbuf[DELIMSZ];
 616         enum mdelim      d;
 617
 618         /* First token must be a close-delimiter. */
 619
 620         for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
 621                 dbuf[j] = buf[i];
 622
 623         if (DELIMSZ == j)
 624                 return(0);
 625
 626         dbuf[j] = '\0';
 627         if (DELIM_CLOSE != mdoc_isdelim(dbuf))
 628                 return(0);
 629
 630         while (' ' == buf[i])
 631                 i++;
 632
 633         /* Remaining must NOT be open/none. */
 634
 635         while (buf[i]) {
 636                 j = 0;
 637                 while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
 638                         dbuf[j++] = buf[i++];
 639
 640                 if (DELIMSZ == j)
 641                         return(0);
 642
 643                 dbuf[j] = '\0';
 644                 d = mdoc_isdelim(dbuf);
 645                 if (DELIM_NONE == d || DELIM_OPEN == d)
 646                         return(0);
 647
 648                 while (' ' == buf[i])
 649                         i++;
 650         }
 651
 652         return('\0' == buf[i]);
 653 }
 654
 655 static int
 656 argv_multi(struct mdoc *mdoc, int line,
 657                 struct mdoc_argv *v, int *pos, char *buf)
 658 {
 659         enum margserr    ac;
 660         char            *p;
 661
 662         for (v->sz = 0; ; v->sz++) {
 663                 if ('-' == buf[*pos])
 664                         break;
 665                 ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
 666                 if (ARGS_ERROR == ac)
 667                         return(0);
 668                 else if (ARGS_EOLN == ac)
 669                         break;
 670
 671                 if (0 == v->sz % MULTI_STEP)
 672                         v->value = mandoc_reallocarray(v->value,
 673                             v->sz + MULTI_STEP, sizeof(char *));
 674
 675                 v->value[(int)v->sz] = mandoc_strdup(p);
 676         }
 677
 678         return(1);
 679 }
 680
 681 static int
 682 argv_single(struct mdoc *mdoc, int line,
 683                 struct mdoc_argv *v, int *pos, char *buf)
 684 {
 685         enum margserr    ac;
 686         char            *p;
 687
 688         ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
 689         if (ARGS_ERROR == ac)
 690                 return(0);
 691         if (ARGS_EOLN == ac)
 692                 return(1);
 693
 694         v->sz = 1;
 695         v->value = mandoc_malloc(sizeof(char *));
 696         v->value[0] = mandoc_strdup(p);
 697
 698         return(1);
 699 }