mdoc_argv.c

   1 /*      $Id: mdoc_argv.c,v 1.60 2010/09/13 20:15:43 schwarze Exp $ */
   2 /*
   3  * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
   4  *
   5  * Permission to use, copy, modify, and distribute this software for any
   6  * purpose with or without fee is hereby granted, provided that the above
   7  * copyright notice and this permission notice appear in all copies.
   8  *
   9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  16  */
  17 #ifdef HAVE_CONFIG_H
  18 #include "config.h"
  19 #endif
  20
  21 #include <sys/types.h>
  22
  23 #include <assert.h>
  24 #include <ctype.h>
  25 #include <stdlib.h>
  26 #include <stdio.h>
  27 #include <string.h>
  28
  29 #include "mandoc.h"
  30 #include "libmdoc.h"
  31 #include "libmandoc.h"
  32
  33 /*
  34  * Routines to parse arguments of macros.  Arguments follow the syntax
  35  * of `-arg [val [valN...]]'.  Arguments come in all types:  quoted
  36  * arguments, multiple arguments per value, no-value arguments, etc.
  37  *
  38  * There's no limit to the number or arguments that may be allocated.
  39  */
  40
  41 #define ARGV_NONE       (1 << 0)
  42 #define ARGV_SINGLE     (1 << 1)
  43 #define ARGV_MULTI      (1 << 2)
  44 #define ARGV_OPT_SINGLE (1 << 3)
  45
  46 #define MULTI_STEP       5
  47
  48 static  enum mdocargt    argv_a2arg(enum mdoct, const char *);
  49 static  enum margserr    args(struct mdoc *, int, int *,
  50                                 char *, int, char **);
  51 static  int              argv(struct mdoc *, int,
  52                                 struct mdoc_argv *, int *, char *);
  53 static  int              argv_single(struct mdoc *, int,
  54                                 struct mdoc_argv *, int *, char *);
  55 static  int              argv_opt_single(struct mdoc *, int,
  56                                 struct mdoc_argv *, int *, char *);
  57 static  int              argv_multi(struct mdoc *, int,
  58                                 struct mdoc_argv *, int *, char *);
  59
  60 /* Per-argument flags. */
  61
  62 static  int mdoc_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_OPT_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  int mdoc_argflags[MDOC_MAX] = {
  92         0, /* Ap */
  93         0, /* Dd */
  94         0, /* Dt */
  95         0, /* Os */
  96         0, /* Sh */
  97         0, /* Ss */
  98         0, /* Pp */
  99         ARGS_DELIM, /* D1 */
 100         ARGS_DELIM, /* Dl */
 101         0, /* Bd */
 102         0, /* Ed */
 103         0, /* Bl */
 104         0, /* El */
 105         0, /* It */
 106         ARGS_DELIM, /* Ad */
 107         ARGS_DELIM, /* An */
 108         ARGS_DELIM, /* Ar */
 109         0, /* Cd */
 110         ARGS_DELIM, /* Cm */
 111         ARGS_DELIM, /* Dv */
 112         ARGS_DELIM, /* Er */
 113         ARGS_DELIM, /* Ev */
 114         0, /* Ex */
 115         ARGS_DELIM, /* Fa */
 116         0, /* Fd */
 117         ARGS_DELIM, /* Fl */
 118         ARGS_DELIM, /* Fn */
 119         ARGS_DELIM, /* Ft */
 120         ARGS_DELIM, /* Ic */
 121         0, /* In */
 122         ARGS_DELIM, /* Li */
 123         0, /* Nd */
 124         ARGS_DELIM, /* Nm */
 125         ARGS_DELIM, /* Op */
 126         0, /* Ot */
 127         ARGS_DELIM, /* Pa */
 128         0, /* Rv */
 129         ARGS_DELIM, /* St */
 130         ARGS_DELIM, /* Va */
 131         ARGS_DELIM, /* Vt */
 132         ARGS_DELIM, /* Xr */
 133         0, /* %A */
 134         0, /* %B */
 135         0, /* %D */
 136         0, /* %I */
 137         0, /* %J */
 138         0, /* %N */
 139         0, /* %O */
 140         0, /* %P */
 141         0, /* %R */
 142         0, /* %T */
 143         0, /* %V */
 144         ARGS_DELIM, /* Ac */
 145         0, /* Ao */
 146         ARGS_DELIM, /* Aq */
 147         ARGS_DELIM, /* At */
 148         ARGS_DELIM, /* Bc */
 149         0, /* Bf */
 150         0, /* Bo */
 151         ARGS_DELIM, /* Bq */
 152         ARGS_DELIM, /* Bsx */
 153         ARGS_DELIM, /* Bx */
 154         0, /* Db */
 155         ARGS_DELIM, /* Dc */
 156         0, /* Do */
 157         ARGS_DELIM, /* Dq */
 158         ARGS_DELIM, /* Ec */
 159         0, /* Ef */
 160         ARGS_DELIM, /* Em */
 161         0, /* Eo */
 162         ARGS_DELIM, /* Fx */
 163         ARGS_DELIM, /* Ms */
 164         ARGS_DELIM, /* No */
 165         ARGS_DELIM, /* Ns */
 166         ARGS_DELIM, /* Nx */
 167         ARGS_DELIM, /* Ox */
 168         ARGS_DELIM, /* Pc */
 169         ARGS_DELIM, /* Pf */
 170         0, /* Po */
 171         ARGS_DELIM, /* Pq */
 172         ARGS_DELIM, /* Qc */
 173         ARGS_DELIM, /* Ql */
 174         0, /* Qo */
 175         ARGS_DELIM, /* Qq */
 176         0, /* Re */
 177         0, /* Rs */
 178         ARGS_DELIM, /* Sc */
 179         0, /* So */
 180         ARGS_DELIM, /* Sq */
 181         0, /* Sm */
 182         ARGS_DELIM, /* Sx */
 183         ARGS_DELIM, /* Sy */
 184         ARGS_DELIM, /* Tn */
 185         ARGS_DELIM, /* Ux */
 186         ARGS_DELIM, /* Xc */
 187         0, /* Xo */
 188         0, /* Fo */
 189         0, /* Fc */
 190         0, /* Oo */
 191         ARGS_DELIM, /* Oc */
 192         0, /* Bk */
 193         0, /* Ek */
 194         0, /* Bt */
 195         0, /* Hf */
 196         0, /* Fr */
 197         0, /* Ud */
 198         0, /* Lb */
 199         0, /* Lp */
 200         ARGS_DELIM, /* Lk */
 201         ARGS_DELIM, /* Mt */
 202         ARGS_DELIM, /* Brq */
 203         0, /* Bro */
 204         ARGS_DELIM, /* Brc */
 205         0, /* %C */
 206         0, /* Es */
 207         0, /* En */
 208         0, /* Dx */
 209         0, /* %Q */
 210         0, /* br */
 211         0, /* sp */
 212         0, /* %U */
 213         0, /* Ta */
 214 };
 215
 216
 217 /*
 218  * Parse an argument from line text.  This comes in the form of -key
 219  * [value0...], which may either have a single mandatory value, at least
 220  * one mandatory value, an optional single value, or no value.
 221  */
 222 enum margverr
 223 mdoc_argv(struct mdoc *m, int line, enum mdoct tok,
 224                 struct mdoc_arg **v, int *pos, char *buf)
 225 {
 226         char             *p, sv;
 227         struct mdoc_argv tmp;
 228         struct mdoc_arg  *arg;
 229
 230         if ('\0' == buf[*pos])
 231                 return(ARGV_EOLN);
 232
 233         assert(' ' != buf[*pos]);
 234
 235         /* Parse through to the first unescaped space. */
 236
 237         p = &buf[++(*pos)];
 238
 239         assert(*pos > 0);
 240
 241         /* LINTED */
 242         while (buf[*pos]) {
 243                 if (' ' == buf[*pos])
 244                         if ('\\' != buf[*pos - 1])
 245                                 break;
 246                 (*pos)++;
 247         }
 248
 249         /* XXX - save zeroed byte, if not an argument. */
 250
 251         sv = '\0';
 252         if (buf[*pos]) {
 253                 sv = buf[*pos];
 254                 buf[(*pos)++] = '\0';
 255         }
 256
 257         (void)memset(&tmp, 0, sizeof(struct mdoc_argv));
 258         tmp.line = line;
 259         tmp.pos = *pos;
 260
 261         /* See if our token accepts the argument. */
 262
 263         if (MDOC_ARG_MAX == (tmp.arg = argv_a2arg(tok, p))) {
 264                 /* XXX - restore saved zeroed byte. */
 265                 if (sv)
 266                         buf[*pos - 1] = sv;
 267                 return(ARGV_WORD);
 268         }
 269
 270         while (buf[*pos] && ' ' == buf[*pos])
 271                 (*pos)++;
 272
 273         if ( ! argv(m, line, &tmp, pos, buf))
 274                 return(ARGV_ERROR);
 275
 276         if (NULL == (arg = *v))
 277                 arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
 278
 279         arg->argc++;
 280         arg->argv = mandoc_realloc
 281                 (arg->argv, arg->argc * sizeof(struct mdoc_argv));
 282
 283         (void)memcpy(&arg->argv[(int)arg->argc - 1],
 284                         &tmp, sizeof(struct mdoc_argv));
 285
 286         return(ARGV_ARG);
 287 }
 288
 289
 290 void
 291 mdoc_argv_free(struct mdoc_arg *p)
 292 {
 293         int              i;
 294
 295         if (NULL == p)
 296                 return;
 297
 298         if (p->refcnt) {
 299                 --(p->refcnt);
 300                 if (p->refcnt)
 301                         return;
 302         }
 303         assert(p->argc);
 304
 305         for (i = (int)p->argc - 1; i >= 0; i--)
 306                 mdoc_argn_free(p, i);
 307
 308         free(p->argv);
 309         free(p);
 310 }
 311
 312
 313 void
 314 mdoc_argn_free(struct mdoc_arg *p, int iarg)
 315 {
 316         struct mdoc_argv *arg;
 317         int               j;
 318
 319         arg = &p->argv[iarg];
 320
 321         if (arg->sz && arg->value) {
 322                 for (j = (int)arg->sz - 1; j >= 0; j--)
 323                         free(arg->value[j]);
 324                 free(arg->value);
 325         }
 326
 327         for (--p->argc; iarg < (int)p->argc; iarg++)
 328                 p->argv[iarg] = p->argv[iarg+1];
 329 }
 330
 331
 332 enum margserr
 333 mdoc_zargs(struct mdoc *m, int line, int *pos,
 334                 char *buf, int flags, char **v)
 335 {
 336
 337         return(args(m, line, pos, buf, flags, v));
 338 }
 339
 340
 341 enum margserr
 342 mdoc_args(struct mdoc *m, int line, int *pos,
 343                 char *buf, enum mdoct tok, char **v)
 344 {
 345         int               fl;
 346         struct mdoc_node *n;
 347
 348         fl = mdoc_argflags[tok];
 349
 350         if (MDOC_It != tok)
 351                 return(args(m, line, pos, buf, fl, v));
 352
 353         /*
 354          * We know that we're in an `It', so it's reasonable to expect
 355          * us to be sitting in a `Bl'.  Someday this may not be the case
 356          * (if we allow random `It's sitting out there), so provide a
 357          * safe fall-back into the default behaviour.
 358          */
 359
 360         for (n = m->last; n; n = n->parent)
 361                 if (MDOC_Bl == n->tok)
 362                         break;
 363
 364         assert(n->data.Bl);
 365         if (n && LIST_column == n->data.Bl->type) {
 366                 fl |= ARGS_TABSEP;
 367                 fl &= ~ARGS_DELIM;
 368         }
 369
 370         return(args(m, line, pos, buf, fl, v));
 371 }
 372
 373
 374 static enum margserr
 375 args(struct mdoc *m, int line, int *pos,
 376                 char *buf, int fl, char **v)
 377 {
 378         int              i;
 379         char            *p, *pp;
 380         enum margserr    rc;
 381         enum mdelim      d;
 382
 383         /*
 384          * Parse out the terms (like `val' in `.Xx -arg val' or simply
 385          * `.Xx val'), which can have all sorts of properties:
 386          *
 387          *   ARGS_DELIM: use special handling if encountering trailing
 388          *   delimiters in the form of [[::delim::][ ]+]+.
 389          *
 390          *   ARGS_NOWARN: don't post warnings.  This is only used when
 391          *   re-parsing delimiters, as the warnings have already been
 392          *   posted.
 393          *
 394          *   ARGS_TABSEP: use special handling for tab/`Ta' separated
 395          *   phrases like in `Bl -column'.
 396          */
 397
 398         assert(' ' != buf[*pos]);
 399
 400         if ('\0' == buf[*pos]) {
 401                 if (MDOC_PPHRASE & m->flags)
 402                         return(ARGS_EOLN);
 403                 /*
 404                  * If we're not in a partial phrase and the flag for
 405                  * being a phrase literal is still set, the punctuation
 406                  * is unterminated.
 407                  */
 408                 if (MDOC_PHRASELIT & m->flags)
 409                         if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE))
 410                                 return(ARGS_ERROR);
 411
 412                 m->flags &= ~MDOC_PHRASELIT;
 413                 return(ARGS_EOLN);
 414         }
 415
 416         /*
 417          * If the first character is a closing delimiter and we're to
 418          * look for delimited strings, then pass down the buffer seeing
 419          * if it follows the pattern of [[::delim::][ ]+]+.  Note that
 420          * we ONLY care about closing delimiters.
 421          */
 422
 423         if ((fl & ARGS_DELIM) && DELIM_CLOSE == mdoc_iscdelim(buf[*pos])) {
 424                 for (i = *pos; buf[i]; ) {
 425                         d = mdoc_iscdelim(buf[i]);
 426                         if (DELIM_NONE == d || DELIM_OPEN == d)
 427                                 break;
 428                         i++;
 429                         if ('\0' == buf[i] || ' ' != buf[i])
 430                                 break;
 431                         i++;
 432                         while (buf[i] && ' ' == buf[i])
 433                                 i++;
 434                 }
 435
 436                 if ('\0' == buf[i]) {
 437                         *v = &buf[*pos];
 438                         if (i && ' ' != buf[i - 1])
 439                                 return(ARGS_PUNCT);
 440                         if (ARGS_NOWARN & fl)
 441                                 return(ARGS_PUNCT);
 442                         if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
 443                                 return(ARGS_ERROR);
 444                         return(ARGS_PUNCT);
 445                 }
 446         }
 447
 448         *v = &buf[*pos];
 449
 450         /*
 451          * First handle TABSEP items, restricted to `Bl -column'.  This
 452          * ignores conventional token parsing and instead uses tabs or
 453          * `Ta' macros to separate phrases.  Phrases are parsed again
 454          * for arguments at a later phase.
 455          */
 456
 457         if (ARGS_TABSEP & fl) {
 458                 /* Scan ahead to tab (can't be escaped). */
 459                 p = strchr(*v, '\t');
 460                 pp = NULL;
 461
 462                 /* Scan ahead to unescaped `Ta'. */
 463                 if ( ! (MDOC_PHRASELIT & m->flags))
 464                         for (pp = *v; ; pp++) {
 465                                 if (NULL == (pp = strstr(pp, "Ta")))
 466                                         break;
 467                                 if (pp > *v && ' ' != *(pp - 1))
 468                                         continue;
 469                                 if (' ' == *(pp + 2) || '\0' == *(pp + 2))
 470                                         break;
 471                         }
 472
 473                 /* By default, assume a phrase. */
 474                 rc = ARGS_PHRASE;
 475
 476                 /*
 477                  * Adjust new-buffer position to be beyond delimiter
 478                  * mark (e.g., Ta -> end + 2).
 479                  */
 480                 if (p && pp) {
 481                         *pos += pp < p ? 2 : 1;
 482                         rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
 483                         p = pp < p ? pp : p;
 484                 } else if (p && ! pp) {
 485                         rc = ARGS_PPHRASE;
 486                         *pos += 1;
 487                 } else if (pp && ! p) {
 488                         p = pp;
 489                         *pos += 2;
 490                 } else {
 491                         rc = ARGS_PEND;
 492                         p = strchr(*v, 0);
 493                 }
 494
 495                 /* Whitespace check for eoln case... */
 496                 if ('\0' == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl))
 497                         if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
 498                                 return(ARGS_ERROR);
 499
 500                 *pos += (int)(p - *v);
 501
 502                 /* Strip delimiter's preceding whitespace. */
 503                 pp = p - 1;
 504                 while (pp > *v && ' ' == *pp) {
 505                         if (pp > *v && '\\' == *(pp - 1))
 506                                 break;
 507                         pp--;
 508                 }
 509                 *(pp + 1) = 0;
 510
 511                 /* Strip delimiter's proceeding whitespace. */
 512                 for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
 513                         /* Skip ahead. */ ;
 514
 515                 return(rc);
 516         }
 517
 518         /*
 519          * Process a quoted literal.  A quote begins with a double-quote
 520          * and ends with a double-quote NOT preceded by a double-quote.
 521          * Whitespace is NOT involved in literal termination.
 522          */
 523
 524         if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
 525                 if ( ! (MDOC_PHRASELIT & m->flags))
 526                         *v = &buf[++(*pos)];
 527
 528                 if (MDOC_PPHRASE & m->flags)
 529                         m->flags |= MDOC_PHRASELIT;
 530
 531                 for ( ; buf[*pos]; (*pos)++) {
 532                         if ('\"' != buf[*pos])
 533                                 continue;
 534                         if ('\"' != buf[*pos + 1])
 535                                 break;
 536                         (*pos)++;
 537                 }
 538
 539                 if ('\0' == buf[*pos]) {
 540                         if (ARGS_NOWARN & fl || MDOC_PPHRASE & m->flags)
 541                                 return(ARGS_QWORD);
 542                         if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE))
 543                                 return(ARGS_ERROR);
 544                         return(ARGS_QWORD);
 545                 }
 546
 547                 m->flags &= ~MDOC_PHRASELIT;
 548                 buf[(*pos)++] = '\0';
 549
 550                 if ('\0' == buf[*pos])
 551                         return(ARGS_QWORD);
 552
 553                 while (' ' == buf[*pos])
 554                         (*pos)++;
 555
 556                 if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
 557                         if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
 558                                 return(ARGS_ERROR);
 559
 560                 return(ARGS_QWORD);
 561         }
 562
 563         /*
 564          * A non-quoted term progresses until either the end of line or
 565          * a non-escaped whitespace.
 566          */
 567
 568         for ( ; buf[*pos]; (*pos)++)
 569                 if (*pos && ' ' == buf[*pos] && '\\' != buf[*pos - 1])
 570                         break;
 571
 572         if ('\0' == buf[*pos])
 573                 return(ARGS_WORD);
 574
 575         buf[(*pos)++] = '\0';
 576
 577         while (' ' == buf[*pos])
 578                 (*pos)++;
 579
 580         if ('\0' == buf[*pos] && ! (ARGS_NOWARN & fl))
 581                 if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
 582                         return(ARGS_ERROR);
 583
 584         return(ARGS_WORD);
 585 }
 586
 587
 588 static enum mdocargt
 589 argv_a2arg(enum mdoct tok, const char *p)
 590 {
 591
 592         /*
 593          * Parse an argument identifier from its text.  XXX - this
 594          * should really be table-driven to clarify the code.
 595          *
 596          * If you add an argument to the list, make sure that you
 597          * register it here with its one or more macros!
 598          */
 599
 600         switch (tok) {
 601         case (MDOC_An):
 602                 if (0 == strcmp(p, "split"))
 603                         return(MDOC_Split);
 604                 else if (0 == strcmp(p, "nosplit"))
 605                         return(MDOC_Nosplit);
 606                 break;
 607
 608         case (MDOC_Bd):
 609                 if (0 == strcmp(p, "ragged"))
 610                         return(MDOC_Ragged);
 611                 else if (0 == strcmp(p, "unfilled"))
 612                         return(MDOC_Unfilled);
 613                 else if (0 == strcmp(p, "filled"))
 614                         return(MDOC_Filled);
 615                 else if (0 == strcmp(p, "literal"))
 616                         return(MDOC_Literal);
 617                 else if (0 == strcmp(p, "file"))
 618                         return(MDOC_File);
 619                 else if (0 == strcmp(p, "offset"))
 620                         return(MDOC_Offset);
 621                 else if (0 == strcmp(p, "compact"))
 622                         return(MDOC_Compact);
 623                 else if (0 == strcmp(p, "centered"))
 624                         return(MDOC_Centred);
 625                 break;
 626
 627         case (MDOC_Bf):
 628                 if (0 == strcmp(p, "emphasis"))
 629                         return(MDOC_Emphasis);
 630                 else if (0 == strcmp(p, "literal"))
 631                         return(MDOC_Literal);
 632                 else if (0 == strcmp(p, "symbolic"))
 633                         return(MDOC_Symbolic);
 634                 break;
 635
 636         case (MDOC_Bk):
 637                 if (0 == strcmp(p, "words"))
 638                         return(MDOC_Words);
 639                 break;
 640
 641         case (MDOC_Bl):
 642                 if (0 == strcmp(p, "bullet"))
 643                         return(MDOC_Bullet);
 644                 else if (0 == strcmp(p, "dash"))
 645                         return(MDOC_Dash);
 646                 else if (0 == strcmp(p, "hyphen"))
 647                         return(MDOC_Hyphen);
 648                 else if (0 == strcmp(p, "item"))
 649                         return(MDOC_Item);
 650                 else if (0 == strcmp(p, "enum"))
 651                         return(MDOC_Enum);
 652                 else if (0 == strcmp(p, "tag"))
 653                         return(MDOC_Tag);
 654                 else if (0 == strcmp(p, "diag"))
 655                         return(MDOC_Diag);
 656                 else if (0 == strcmp(p, "hang"))
 657                         return(MDOC_Hang);
 658                 else if (0 == strcmp(p, "ohang"))
 659                         return(MDOC_Ohang);
 660                 else if (0 == strcmp(p, "inset"))
 661                         return(MDOC_Inset);
 662                 else if (0 == strcmp(p, "column"))
 663                         return(MDOC_Column);
 664                 else if (0 == strcmp(p, "width"))
 665                         return(MDOC_Width);
 666                 else if (0 == strcmp(p, "offset"))
 667                         return(MDOC_Offset);
 668                 else if (0 == strcmp(p, "compact"))
 669                         return(MDOC_Compact);
 670                 else if (0 == strcmp(p, "nested"))
 671                         return(MDOC_Nested);
 672                 break;
 673
 674         case (MDOC_Rv):
 675                 /* FALLTHROUGH */
 676         case (MDOC_Ex):
 677                 if (0 == strcmp(p, "std"))
 678                         return(MDOC_Std);
 679                 break;
 680         default:
 681                 break;
 682         }
 683
 684         return(MDOC_ARG_MAX);
 685 }
 686
 687
 688 static int
 689 argv_multi(struct mdoc *m, int line,
 690                 struct mdoc_argv *v, int *pos, char *buf)
 691 {
 692         enum margserr    ac;
 693         char            *p;
 694
 695         for (v->sz = 0; ; v->sz++) {
 696                 if ('-' == buf[*pos])
 697                         break;
 698                 ac = args(m, line, pos, buf, 0, &p);
 699                 if (ARGS_ERROR == ac)
 700                         return(0);
 701                 else if (ARGS_EOLN == ac)
 702                         break;
 703
 704                 if (0 == v->sz % MULTI_STEP)
 705                         v->value = mandoc_realloc(v->value,
 706                                 (v->sz + MULTI_STEP) * sizeof(char *));
 707
 708                 v->value[(int)v->sz] = mandoc_strdup(p);
 709         }
 710
 711         return(1);
 712 }
 713
 714
 715 static int
 716 argv_opt_single(struct mdoc *m, int line,
 717                 struct mdoc_argv *v, int *pos, char *buf)
 718 {
 719         enum margserr    ac;
 720         char            *p;
 721
 722         if ('-' == buf[*pos])
 723                 return(1);
 724
 725         ac = args(m, line, pos, buf, 0, &p);
 726         if (ARGS_ERROR == ac)
 727                 return(0);
 728         if (ARGS_EOLN == ac)
 729                 return(1);
 730
 731         v->sz = 1;
 732         v->value = mandoc_malloc(sizeof(char *));
 733         v->value[0] = mandoc_strdup(p);
 734
 735         return(1);
 736 }
 737
 738
 739 /*
 740  * Parse a single, mandatory value from the stream.
 741  */
 742 static int
 743 argv_single(struct mdoc *m, int line,
 744                 struct mdoc_argv *v, int *pos, char *buf)
 745 {
 746         int              ppos;
 747         enum margserr    ac;
 748         char            *p;
 749
 750         ppos = *pos;
 751
 752         ac = args(m, line, pos, buf, 0, &p);
 753         if (ARGS_EOLN == ac) {
 754                 mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTARGVCOUNT);
 755                 return(0);
 756         } else if (ARGS_ERROR == ac)
 757                 return(0);
 758
 759         v->sz = 1;
 760         v->value = mandoc_malloc(sizeof(char *));
 761         v->value[0] = mandoc_strdup(p);
 762
 763         return(1);
 764 }
 765
 766
 767 /*
 768  * Determine rules for parsing arguments.  Arguments can either accept
 769  * no parameters, an optional single parameter, one parameter, or
 770  * multiple parameters.
 771  */
 772 static int
 773 argv(struct mdoc *mdoc, int line,
 774                 struct mdoc_argv *v, int *pos, char *buf)
 775 {
 776
 777         v->sz = 0;
 778         v->value = NULL;
 779
 780         switch (mdoc_argvflags[v->arg]) {
 781         case (ARGV_SINGLE):
 782                 return(argv_single(mdoc, line, v, pos, buf));
 783         case (ARGV_MULTI):
 784                 return(argv_multi(mdoc, line, v, pos, buf));
 785         case (ARGV_OPT_SINGLE):
 786                 return(argv_opt_single(mdoc, line, v, pos, buf));
 787         default:
 788                 /* ARGV_NONE */
 789                 break;
 790         }
 791
 792         return(1);
 793 }