Annotation of mandoc/mdoc_argv.c, Revision 1.78
1.78 ! kristaps 1: /* $Id: mdoc_argv.c,v 1.77 2011/05/12 23:44:01 kristaps Exp $ */
1.1 kristaps 2: /*
1.56 schwarze 3: * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons <kristaps@bsd.lv>
1.1 kristaps 4: *
5: * Permission to use, copy, modify, and distribute this software for any
1.3 kristaps 6: * purpose with or without fee is hereby granted, provided that the above
7: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 8: *
1.3 kristaps 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.
1.1 kristaps 16: */
1.33 kristaps 17: #ifdef HAVE_CONFIG_H
18: #include "config.h"
19: #endif
20:
1.1 kristaps 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:
1.72 kristaps 29: #include "mdoc.h"
1.49 kristaps 30: #include "mandoc.h"
1.1 kristaps 31: #include "libmdoc.h"
1.31 kristaps 32: #include "libmandoc.h"
1.1 kristaps 33:
1.68 kristaps 34: #define MULTI_STEP 5 /* pre-allocate argument values */
1.76 kristaps 35: #define DELIMSZ 6 /* max possible size of a delimiter */
36:
37: enum argsflag {
38: ARGSFL_NONE = 0,
39: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
40: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
41: };
42:
43: enum argvflag {
44: ARGV_NONE, /* no args to flag (e.g., -split) */
45: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
46: ARGV_MULTI, /* multiple args (e.g., -column xxx yyy) */
47: ARGV_OPT_SINGLE /* optional arg (e.g., -offset [xxx]) */
48: };
1.1 kristaps 49:
1.78 ! kristaps 50: struct mdocarg {
! 51: enum argsflag flags;
! 52: const enum mdocargt *argvs;
! 53: };
! 54:
1.39 kristaps 55: static enum margserr args(struct mdoc *, int, int *,
1.76 kristaps 56: char *, enum argsflag, char **);
1.77 kristaps 57: static int args_checkpunct(const char *, int);
1.1 kristaps 58: static int argv(struct mdoc *, int,
59: struct mdoc_argv *, int *, char *);
60: static int argv_single(struct mdoc *, int,
61: struct mdoc_argv *, int *, char *);
62: static int argv_opt_single(struct mdoc *, int,
63: struct mdoc_argv *, int *, char *);
64: static int argv_multi(struct mdoc *, int,
65: struct mdoc_argv *, int *, char *);
1.71 kristaps 66: static void argn_free(struct mdoc_arg *, int);
1.2 kristaps 67:
1.68 kristaps 68: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 69: ARGV_NONE, /* MDOC_Split */
70: ARGV_NONE, /* MDOC_Nosplit */
71: ARGV_NONE, /* MDOC_Ragged */
72: ARGV_NONE, /* MDOC_Unfilled */
73: ARGV_NONE, /* MDOC_Literal */
1.29 kristaps 74: ARGV_SINGLE, /* MDOC_File */
1.26 kristaps 75: ARGV_OPT_SINGLE, /* MDOC_Offset */
1.1 kristaps 76: ARGV_NONE, /* MDOC_Bullet */
77: ARGV_NONE, /* MDOC_Dash */
78: ARGV_NONE, /* MDOC_Hyphen */
79: ARGV_NONE, /* MDOC_Item */
80: ARGV_NONE, /* MDOC_Enum */
81: ARGV_NONE, /* MDOC_Tag */
82: ARGV_NONE, /* MDOC_Diag */
83: ARGV_NONE, /* MDOC_Hang */
84: ARGV_NONE, /* MDOC_Ohang */
85: ARGV_NONE, /* MDOC_Inset */
86: ARGV_MULTI, /* MDOC_Column */
87: ARGV_SINGLE, /* MDOC_Width */
88: ARGV_NONE, /* MDOC_Compact */
1.25 kristaps 89: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 90: ARGV_NONE, /* MDOC_Filled */
91: ARGV_NONE, /* MDOC_Words */
92: ARGV_NONE, /* MDOC_Emphasis */
93: ARGV_NONE, /* MDOC_Symbolic */
94: ARGV_NONE /* MDOC_Symbolic */
95: };
96:
1.73 kristaps 97: static const enum mdocargt args_Ex[] = {
98: MDOC_Std,
99: MDOC_ARG_MAX
100: };
101:
102: static const enum mdocargt args_An[] = {
103: MDOC_Split,
104: MDOC_Nosplit,
105: MDOC_ARG_MAX
106: };
107:
108: static const enum mdocargt args_Bd[] = {
109: MDOC_Ragged,
110: MDOC_Unfilled,
111: MDOC_Filled,
112: MDOC_Literal,
113: MDOC_File,
114: MDOC_Offset,
115: MDOC_Compact,
116: MDOC_Centred,
117: MDOC_ARG_MAX
118: };
119:
120: static const enum mdocargt args_Bf[] = {
121: MDOC_Emphasis,
122: MDOC_Literal,
123: MDOC_Symbolic,
124: MDOC_ARG_MAX
125: };
126:
127: static const enum mdocargt args_Bk[] = {
128: MDOC_Words,
129: MDOC_ARG_MAX
130: };
131:
132: static const enum mdocargt args_Bl[] = {
133: MDOC_Bullet,
134: MDOC_Dash,
135: MDOC_Hyphen,
136: MDOC_Item,
137: MDOC_Enum,
138: MDOC_Tag,
139: MDOC_Diag,
140: MDOC_Hang,
141: MDOC_Ohang,
142: MDOC_Inset,
143: MDOC_Column,
144: MDOC_Width,
145: MDOC_Offset,
146: MDOC_Compact,
147: MDOC_Nested,
148: MDOC_ARG_MAX
149: };
150:
1.78 ! kristaps 151: static const struct mdocarg mdocargs[MDOC_MAX] = {
! 152: { ARGSFL_NONE, NULL }, /* Ap */
! 153: { ARGSFL_NONE, NULL }, /* Dd */
! 154: { ARGSFL_NONE, NULL }, /* Dt */
! 155: { ARGSFL_NONE, NULL }, /* Os */
! 156: { ARGSFL_NONE, NULL }, /* Sh */
! 157: { ARGSFL_NONE, NULL }, /* Ss */
! 158: { ARGSFL_NONE, NULL }, /* Pp */
! 159: { ARGSFL_DELIM, NULL }, /* D1 */
! 160: { ARGSFL_DELIM, NULL }, /* Dl */
! 161: { ARGSFL_NONE, args_Bd }, /* Bd */
! 162: { ARGSFL_NONE, NULL }, /* Ed */
! 163: { ARGSFL_NONE, args_Bl }, /* Bl */
! 164: { ARGSFL_NONE, NULL }, /* El */
! 165: { ARGSFL_NONE, NULL }, /* It */
! 166: { ARGSFL_DELIM, NULL }, /* Ad */
! 167: { ARGSFL_DELIM, args_An }, /* An */
! 168: { ARGSFL_DELIM, NULL }, /* Ar */
! 169: { ARGSFL_NONE, NULL }, /* Cd */
! 170: { ARGSFL_DELIM, NULL }, /* Cm */
! 171: { ARGSFL_DELIM, NULL }, /* Dv */
! 172: { ARGSFL_DELIM, NULL }, /* Er */
! 173: { ARGSFL_DELIM, NULL }, /* Ev */
! 174: { ARGSFL_NONE, args_Ex }, /* Ex */
! 175: { ARGSFL_DELIM, NULL }, /* Fa */
! 176: { ARGSFL_NONE, NULL }, /* Fd */
! 177: { ARGSFL_DELIM, NULL }, /* Fl */
! 178: { ARGSFL_DELIM, NULL }, /* Fn */
! 179: { ARGSFL_DELIM, NULL }, /* Ft */
! 180: { ARGSFL_DELIM, NULL }, /* Ic */
! 181: { ARGSFL_NONE, NULL }, /* In */
! 182: { ARGSFL_DELIM, NULL }, /* Li */
! 183: { ARGSFL_NONE, NULL }, /* Nd */
! 184: { ARGSFL_DELIM, NULL }, /* Nm */
! 185: { ARGSFL_DELIM, NULL }, /* Op */
! 186: { ARGSFL_NONE, NULL }, /* Ot */
! 187: { ARGSFL_DELIM, NULL }, /* Pa */
! 188: { ARGSFL_NONE, args_Ex }, /* Rv */
! 189: { ARGSFL_DELIM, NULL }, /* St */
! 190: { ARGSFL_DELIM, NULL }, /* Va */
! 191: { ARGSFL_DELIM, NULL }, /* Vt */
! 192: { ARGSFL_DELIM, NULL }, /* Xr */
! 193: { ARGSFL_NONE, NULL }, /* %A */
! 194: { ARGSFL_NONE, NULL }, /* %B */
! 195: { ARGSFL_NONE, NULL }, /* %D */
! 196: { ARGSFL_NONE, NULL }, /* %I */
! 197: { ARGSFL_NONE, NULL }, /* %J */
! 198: { ARGSFL_NONE, NULL }, /* %N */
! 199: { ARGSFL_NONE, NULL }, /* %O */
! 200: { ARGSFL_NONE, NULL }, /* %P */
! 201: { ARGSFL_NONE, NULL }, /* %R */
! 202: { ARGSFL_NONE, NULL }, /* %T */
! 203: { ARGSFL_NONE, NULL }, /* %V */
! 204: { ARGSFL_DELIM, NULL }, /* Ac */
! 205: { ARGSFL_NONE, NULL }, /* Ao */
! 206: { ARGSFL_DELIM, NULL }, /* Aq */
! 207: { ARGSFL_DELIM, NULL }, /* At */
! 208: { ARGSFL_DELIM, NULL }, /* Bc */
! 209: { ARGSFL_NONE, args_Bf }, /* Bf */
! 210: { ARGSFL_NONE, NULL }, /* Bo */
! 211: { ARGSFL_DELIM, NULL }, /* Bq */
! 212: { ARGSFL_DELIM, NULL }, /* Bsx */
! 213: { ARGSFL_DELIM, NULL }, /* Bx */
! 214: { ARGSFL_NONE, NULL }, /* Db */
! 215: { ARGSFL_DELIM, NULL }, /* Dc */
! 216: { ARGSFL_NONE, NULL }, /* Do */
! 217: { ARGSFL_DELIM, NULL }, /* Dq */
! 218: { ARGSFL_DELIM, NULL }, /* Ec */
! 219: { ARGSFL_NONE, NULL }, /* Ef */
! 220: { ARGSFL_DELIM, NULL }, /* Em */
! 221: { ARGSFL_NONE, NULL }, /* Eo */
! 222: { ARGSFL_DELIM, NULL }, /* Fx */
! 223: { ARGSFL_DELIM, NULL }, /* Ms */
! 224: { ARGSFL_DELIM, NULL }, /* No */
! 225: { ARGSFL_DELIM, NULL }, /* Ns */
! 226: { ARGSFL_DELIM, NULL }, /* Nx */
! 227: { ARGSFL_DELIM, NULL }, /* Ox */
! 228: { ARGSFL_DELIM, NULL }, /* Pc */
! 229: { ARGSFL_DELIM, NULL }, /* Pf */
! 230: { ARGSFL_NONE, NULL }, /* Po */
! 231: { ARGSFL_DELIM, NULL }, /* Pq */
! 232: { ARGSFL_DELIM, NULL }, /* Qc */
! 233: { ARGSFL_DELIM, NULL }, /* Ql */
! 234: { ARGSFL_NONE, NULL }, /* Qo */
! 235: { ARGSFL_DELIM, NULL }, /* Qq */
! 236: { ARGSFL_NONE, NULL }, /* Re */
! 237: { ARGSFL_NONE, NULL }, /* Rs */
! 238: { ARGSFL_DELIM, NULL }, /* Sc */
! 239: { ARGSFL_NONE, NULL }, /* So */
! 240: { ARGSFL_DELIM, NULL }, /* Sq */
! 241: { ARGSFL_NONE, NULL }, /* Sm */
! 242: { ARGSFL_DELIM, NULL }, /* Sx */
! 243: { ARGSFL_DELIM, NULL }, /* Sy */
! 244: { ARGSFL_DELIM, NULL }, /* Tn */
! 245: { ARGSFL_DELIM, NULL }, /* Ux */
! 246: { ARGSFL_DELIM, NULL }, /* Xc */
! 247: { ARGSFL_NONE, NULL }, /* Xo */
! 248: { ARGSFL_NONE, NULL }, /* Fo */
! 249: { ARGSFL_NONE, NULL }, /* Fc */
! 250: { ARGSFL_NONE, NULL }, /* Oo */
! 251: { ARGSFL_DELIM, NULL }, /* Oc */
! 252: { ARGSFL_NONE, args_Bk }, /* Bk */
! 253: { ARGSFL_NONE, NULL }, /* Ek */
! 254: { ARGSFL_NONE, NULL }, /* Bt */
! 255: { ARGSFL_NONE, NULL }, /* Hf */
! 256: { ARGSFL_NONE, NULL }, /* Fr */
! 257: { ARGSFL_NONE, NULL }, /* Ud */
! 258: { ARGSFL_NONE, NULL }, /* Lb */
! 259: { ARGSFL_NONE, NULL }, /* Lp */
! 260: { ARGSFL_DELIM, NULL }, /* Lk */
! 261: { ARGSFL_DELIM, NULL }, /* Mt */
! 262: { ARGSFL_DELIM, NULL }, /* Brq */
! 263: { ARGSFL_NONE, NULL }, /* Bro */
! 264: { ARGSFL_DELIM, NULL }, /* Brc */
! 265: { ARGSFL_NONE, NULL }, /* %C */
! 266: { ARGSFL_NONE, NULL }, /* Es */
! 267: { ARGSFL_NONE, NULL }, /* En */
! 268: { ARGSFL_NONE, NULL }, /* Dx */
! 269: { ARGSFL_NONE, NULL }, /* %Q */
! 270: { ARGSFL_NONE, NULL }, /* br */
! 271: { ARGSFL_NONE, NULL }, /* sp */
! 272: { ARGSFL_NONE, NULL }, /* %U */
! 273: { ARGSFL_NONE, NULL }, /* Ta */
! 274: };
! 275:
! 276:
1.1 kristaps 277: /*
278: * Parse an argument from line text. This comes in the form of -key
279: * [value0...], which may either have a single mandatory value, at least
280: * one mandatory value, an optional single value, or no value.
281: */
1.40 kristaps 282: enum margverr
1.35 kristaps 283: mdoc_argv(struct mdoc *m, int line, enum mdoct tok,
1.1 kristaps 284: struct mdoc_arg **v, int *pos, char *buf)
285: {
286: char *p, sv;
287: struct mdoc_argv tmp;
288: struct mdoc_arg *arg;
1.78 ! kristaps 289: const enum mdocargt *ap;
1.1 kristaps 290:
1.52 kristaps 291: if ('\0' == buf[*pos])
1.1 kristaps 292: return(ARGV_EOLN);
1.78 ! kristaps 293: else if (NULL == (ap = mdocargs[tok].argvs))
! 294: return(ARGV_WORD);
1.1 kristaps 295:
296: assert(' ' != buf[*pos]);
297:
1.78 ! kristaps 298: /* Seek to the first unescaped space. */
1.1 kristaps 299:
300: p = &buf[++(*pos)];
301:
302: assert(*pos > 0);
303:
1.78 ! kristaps 304: for ( ; buf[*pos] ; (*pos)++)
! 305: if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
! 306: break;
1.1 kristaps 307:
1.78 ! kristaps 308: /*
! 309: * We want to nil-terminate the word to look it up (it's easier
! 310: * that way). But we may not have a flag, in which case we need
! 311: * to restore the line as-is. So keep around the stray byte,
! 312: * which we'll reset upon exiting (if necessary).
! 313: */
1.1 kristaps 314:
1.78 ! kristaps 315: if ('\0' != (sv = buf[*pos]))
1.52 kristaps 316: buf[(*pos)++] = '\0';
1.78 ! kristaps 317:
! 318: /*
! 319: * Now look up the word as a flag. Use temporary storage that
! 320: * we'll copy into the node's flags, if necessary.
! 321: */
1.1 kristaps 322:
1.68 kristaps 323: memset(&tmp, 0, sizeof(struct mdoc_argv));
1.78 ! kristaps 324:
1.1 kristaps 325: tmp.line = line;
326: tmp.pos = *pos;
1.78 ! kristaps 327: tmp.arg = MDOC_ARG_MAX;
1.1 kristaps 328:
1.78 ! kristaps 329: while (MDOC_ARG_MAX != (tmp.arg = *ap++))
! 330: if (0 == strcmp(p, mdoc_argnames[tmp.arg]))
! 331: break;
1.1 kristaps 332:
1.78 ! kristaps 333: if (MDOC_ARG_MAX == tmp.arg) {
! 334: /*
! 335: * The flag was not found.
! 336: * Restore saved zeroed byte and return as a word.
! 337: */
1.1 kristaps 338: if (sv)
339: buf[*pos - 1] = sv;
340: return(ARGV_WORD);
341: }
342:
1.78 ! kristaps 343: /* Read to the next word (the argument). */
! 344:
1.1 kristaps 345: while (buf[*pos] && ' ' == buf[*pos])
346: (*pos)++;
347:
1.11 kristaps 348: if ( ! argv(m, line, &tmp, pos, buf))
1.1 kristaps 349: return(ARGV_ERROR);
350:
1.31 kristaps 351: if (NULL == (arg = *v))
1.32 kristaps 352: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 353:
354: arg->argc++;
1.31 kristaps 355: arg->argv = mandoc_realloc
356: (arg->argv, arg->argc * sizeof(struct mdoc_argv));
1.2 kristaps 357:
1.68 kristaps 358: memcpy(&arg->argv[(int)arg->argc - 1],
1.1 kristaps 359: &tmp, sizeof(struct mdoc_argv));
360:
361: return(ARGV_ARG);
362: }
363:
364: void
365: mdoc_argv_free(struct mdoc_arg *p)
366: {
1.37 kristaps 367: int i;
1.1 kristaps 368:
369: if (NULL == p)
370: return;
371:
372: if (p->refcnt) {
373: --(p->refcnt);
374: if (p->refcnt)
375: return;
376: }
377: assert(p->argc);
378:
1.37 kristaps 379: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 380: argn_free(p, i);
1.37 kristaps 381:
382: free(p->argv);
383: free(p);
384: }
385:
1.71 kristaps 386: static void
387: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 388: {
1.58 kristaps 389: struct mdoc_argv *arg;
1.37 kristaps 390: int j;
1.58 kristaps 391:
392: arg = &p->argv[iarg];
1.1 kristaps 393:
1.37 kristaps 394: if (arg->sz && arg->value) {
395: for (j = (int)arg->sz - 1; j >= 0; j--)
396: free(arg->value[j]);
397: free(arg->value);
1.1 kristaps 398: }
399:
1.37 kristaps 400: for (--p->argc; iarg < (int)p->argc; iarg++)
401: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 402: }
403:
1.39 kristaps 404: enum margserr
1.76 kristaps 405: mdoc_zargs(struct mdoc *m, int line, int *pos, char *buf, char **v)
1.19 kristaps 406: {
407:
1.76 kristaps 408: return(args(m, line, pos, buf, ARGSFL_NONE, v));
1.19 kristaps 409: }
410:
1.39 kristaps 411: enum margserr
1.35 kristaps 412: mdoc_args(struct mdoc *m, int line, int *pos,
413: char *buf, enum mdoct tok, char **v)
1.1 kristaps 414: {
1.76 kristaps 415: enum argsflag fl;
1.1 kristaps 416: struct mdoc_node *n;
417:
1.78 ! kristaps 418: fl = mdocargs[tok].flags;
1.1 kristaps 419:
1.17 kristaps 420: if (MDOC_It != tok)
421: return(args(m, line, pos, buf, fl, v));
422:
1.51 kristaps 423: /*
424: * We know that we're in an `It', so it's reasonable to expect
425: * us to be sitting in a `Bl'. Someday this may not be the case
426: * (if we allow random `It's sitting out there), so provide a
427: * safe fall-back into the default behaviour.
428: */
429:
1.17 kristaps 430: for (n = m->last; n; n = n->parent)
1.50 kristaps 431: if (MDOC_Bl == n->tok)
1.76 kristaps 432: if (LIST_column == n->norm->Bl.type) {
433: fl = ARGSFL_TABSEP;
434: break;
435: }
1.1 kristaps 436:
1.11 kristaps 437: return(args(m, line, pos, buf, fl, v));
1.1 kristaps 438: }
439:
1.39 kristaps 440: static enum margserr
1.13 kristaps 441: args(struct mdoc *m, int line, int *pos,
1.76 kristaps 442: char *buf, enum argsflag fl, char **v)
1.1 kristaps 443: {
1.52 kristaps 444: char *p, *pp;
445: enum margserr rc;
1.1 kristaps 446:
1.46 kristaps 447: if ('\0' == buf[*pos]) {
1.47 kristaps 448: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 449: return(ARGS_EOLN);
450: /*
451: * If we're not in a partial phrase and the flag for
452: * being a phrase literal is still set, the punctuation
453: * is unterminated.
454: */
455: if (MDOC_PHRASELIT & m->flags)
1.67 kristaps 456: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.46 kristaps 457:
458: m->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 459: return(ARGS_EOLN);
1.46 kristaps 460: }
1.1 kristaps 461:
1.64 kristaps 462: *v = &buf[*pos];
1.17 kristaps 463:
1.76 kristaps 464: if (ARGSFL_DELIM == fl)
1.77 kristaps 465: if (args_checkpunct(buf, *pos))
1.76 kristaps 466: return(ARGS_PUNCT);
1.1 kristaps 467:
1.17 kristaps 468: /*
469: * First handle TABSEP items, restricted to `Bl -column'. This
470: * ignores conventional token parsing and instead uses tabs or
471: * `Ta' macros to separate phrases. Phrases are parsed again
472: * for arguments at a later phase.
473: */
1.1 kristaps 474:
1.76 kristaps 475: if (ARGSFL_TABSEP == fl) {
1.20 kristaps 476: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 477: p = strchr(*v, '\t');
1.46 kristaps 478: pp = NULL;
1.17 kristaps 479:
480: /* Scan ahead to unescaped `Ta'. */
1.46 kristaps 481: if ( ! (MDOC_PHRASELIT & m->flags))
482: for (pp = *v; ; pp++) {
483: if (NULL == (pp = strstr(pp, "Ta")))
484: break;
485: if (pp > *v && ' ' != *(pp - 1))
486: continue;
1.51 kristaps 487: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 488: break;
489: }
1.17 kristaps 490:
1.42 kristaps 491: /* By default, assume a phrase. */
492: rc = ARGS_PHRASE;
493:
1.21 kristaps 494: /*
495: * Adjust new-buffer position to be beyond delimiter
496: * mark (e.g., Ta -> end + 2).
497: */
1.20 kristaps 498: if (p && pp) {
499: *pos += pp < p ? 2 : 1;
1.42 kristaps 500: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 501: p = pp < p ? pp : p;
502: } else if (p && ! pp) {
1.42 kristaps 503: rc = ARGS_PPHRASE;
1.20 kristaps 504: *pos += 1;
505: } else if (pp && ! p) {
1.17 kristaps 506: p = pp;
1.20 kristaps 507: *pos += 2;
1.44 kristaps 508: } else {
509: rc = ARGS_PEND;
1.20 kristaps 510: p = strchr(*v, 0);
1.44 kristaps 511: }
1.20 kristaps 512:
1.21 kristaps 513: /* Whitespace check for eoln case... */
1.76 kristaps 514: if ('\0' == *p && ' ' == *(p - 1))
1.67 kristaps 515: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.20 kristaps 516:
517: *pos += (int)(p - *v);
1.17 kristaps 518:
519: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 520: pp = p - 1;
521: while (pp > *v && ' ' == *pp) {
522: if (pp > *v && '\\' == *(pp - 1))
523: break;
524: pp--;
1.17 kristaps 525: }
1.20 kristaps 526: *(pp + 1) = 0;
1.17 kristaps 527:
1.20 kristaps 528: /* Strip delimiter's proceeding whitespace. */
529: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
530: /* Skip ahead. */ ;
1.1 kristaps 531:
1.41 kristaps 532: return(rc);
1.17 kristaps 533: }
1.1 kristaps 534:
1.17 kristaps 535: /*
536: * Process a quoted literal. A quote begins with a double-quote
537: * and ends with a double-quote NOT preceded by a double-quote.
538: * Whitespace is NOT involved in literal termination.
539: */
1.1 kristaps 540:
1.46 kristaps 541: if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
542: if ( ! (MDOC_PHRASELIT & m->flags))
543: *v = &buf[++(*pos)];
544:
1.47 kristaps 545: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 546: m->flags |= MDOC_PHRASELIT;
1.1 kristaps 547:
1.17 kristaps 548: for ( ; buf[*pos]; (*pos)++) {
549: if ('\"' != buf[*pos])
550: continue;
551: if ('\"' != buf[*pos + 1])
552: break;
553: (*pos)++;
554: }
1.1 kristaps 555:
1.46 kristaps 556: if ('\0' == buf[*pos]) {
1.76 kristaps 557: if (MDOC_PPHRASE & m->flags)
1.23 kristaps 558: return(ARGS_QWORD);
1.67 kristaps 559: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.18 kristaps 560: return(ARGS_QWORD);
1.17 kristaps 561: }
1.1 kristaps 562:
1.46 kristaps 563: m->flags &= ~MDOC_PHRASELIT;
564: buf[(*pos)++] = '\0';
1.1 kristaps 565:
1.46 kristaps 566: if ('\0' == buf[*pos])
1.17 kristaps 567: return(ARGS_QWORD);
1.1 kristaps 568:
1.17 kristaps 569: while (' ' == buf[*pos])
570: (*pos)++;
1.1 kristaps 571:
1.76 kristaps 572: if ('\0' == buf[*pos])
1.67 kristaps 573: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 574:
1.17 kristaps 575: return(ARGS_QWORD);
1.1 kristaps 576: }
577:
1.75 kristaps 578: p = &buf[*pos];
1.76 kristaps 579: *v = mandoc_getarg(m->parse, &p, line, pos);
1.1 kristaps 580:
1.17 kristaps 581: return(ARGS_WORD);
1.64 kristaps 582: }
583:
584: /*
585: * Check if the string consists only of space-separated closing
1.65 kristaps 586: * delimiters. This is a bit of a dance: the first must be a close
587: * delimiter, but it may be followed by middle delimiters. Arbitrary
588: * whitespace may separate these tokens.
1.64 kristaps 589: */
590: static int
1.77 kristaps 591: args_checkpunct(const char *buf, int i)
1.64 kristaps 592: {
1.69 kristaps 593: int j;
594: char dbuf[DELIMSZ];
1.64 kristaps 595: enum mdelim d;
596:
1.65 kristaps 597: /* First token must be a close-delimiter. */
598:
1.69 kristaps 599: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
600: dbuf[j] = buf[i];
1.65 kristaps 601:
602: if (DELIMSZ == j)
603: return(0);
604:
1.69 kristaps 605: dbuf[j] = '\0';
1.70 kristaps 606: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.64 kristaps 607: return(0);
608:
1.69 kristaps 609: while (' ' == buf[i])
1.65 kristaps 610: i++;
611:
612: /* Remaining must NOT be open/none. */
613:
1.69 kristaps 614: while (buf[i]) {
1.65 kristaps 615: j = 0;
1.69 kristaps 616: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
617: dbuf[j++] = buf[i++];
1.65 kristaps 618:
619: if (DELIMSZ == j)
620: return(0);
621:
1.69 kristaps 622: dbuf[j] = '\0';
1.70 kristaps 623: d = mdoc_isdelim(dbuf);
1.64 kristaps 624: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 625: return(0);
626:
1.69 kristaps 627: while (' ' == buf[i])
1.64 kristaps 628: i++;
629: }
1.69 kristaps 630:
631: return('\0' == buf[i]);
1.1 kristaps 632: }
633:
634: static int
1.11 kristaps 635: argv_multi(struct mdoc *m, int line,
1.1 kristaps 636: struct mdoc_argv *v, int *pos, char *buf)
637: {
1.43 kristaps 638: enum margserr ac;
1.1 kristaps 639: char *p;
640:
641: for (v->sz = 0; ; v->sz++) {
642: if ('-' == buf[*pos])
643: break;
1.77 kristaps 644: ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 645: if (ARGS_ERROR == ac)
1.1 kristaps 646: return(0);
1.43 kristaps 647: else if (ARGS_EOLN == ac)
1.1 kristaps 648: break;
649:
1.31 kristaps 650: if (0 == v->sz % MULTI_STEP)
651: v->value = mandoc_realloc(v->value,
1.1 kristaps 652: (v->sz + MULTI_STEP) * sizeof(char *));
1.31 kristaps 653:
654: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 655: }
656:
1.7 kristaps 657: return(1);
1.1 kristaps 658: }
659:
660: static int
1.11 kristaps 661: argv_opt_single(struct mdoc *m, int line,
1.1 kristaps 662: struct mdoc_argv *v, int *pos, char *buf)
663: {
1.43 kristaps 664: enum margserr ac;
1.1 kristaps 665: char *p;
666:
667: if ('-' == buf[*pos])
668: return(1);
669:
1.77 kristaps 670: ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 671: if (ARGS_ERROR == ac)
1.1 kristaps 672: return(0);
1.43 kristaps 673: if (ARGS_EOLN == ac)
1.1 kristaps 674: return(1);
675:
676: v->sz = 1;
1.31 kristaps 677: v->value = mandoc_malloc(sizeof(char *));
678: v->value[0] = mandoc_strdup(p);
1.2 kristaps 679:
1.1 kristaps 680: return(1);
681: }
682:
683: /*
684: * Parse a single, mandatory value from the stream.
685: */
686: static int
1.11 kristaps 687: argv_single(struct mdoc *m, int line,
1.1 kristaps 688: struct mdoc_argv *v, int *pos, char *buf)
689: {
1.43 kristaps 690: int ppos;
691: enum margserr ac;
1.1 kristaps 692: char *p;
693:
694: ppos = *pos;
695:
1.77 kristaps 696: ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
1.49 kristaps 697: if (ARGS_EOLN == ac) {
698: mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTARGVCOUNT);
699: return(0);
700: } else if (ARGS_ERROR == ac)
1.1 kristaps 701: return(0);
702:
703: v->sz = 1;
1.31 kristaps 704: v->value = mandoc_malloc(sizeof(char *));
705: v->value[0] = mandoc_strdup(p);
1.2 kristaps 706:
1.1 kristaps 707: return(1);
708: }
709:
710: /*
711: * Determine rules for parsing arguments. Arguments can either accept
712: * no parameters, an optional single parameter, one parameter, or
713: * multiple parameters.
714: */
715: static int
716: argv(struct mdoc *mdoc, int line,
717: struct mdoc_argv *v, int *pos, char *buf)
718: {
719:
720: v->sz = 0;
721: v->value = NULL;
722:
1.68 kristaps 723: switch (argvflags[v->arg]) {
1.1 kristaps 724: case (ARGV_SINGLE):
725: return(argv_single(mdoc, line, v, pos, buf));
726: case (ARGV_MULTI):
727: return(argv_multi(mdoc, line, v, pos, buf));
728: case (ARGV_OPT_SINGLE):
729: return(argv_opt_single(mdoc, line, v, pos, buf));
1.68 kristaps 730: case (ARGV_NONE):
731: break;
1.1 kristaps 732: default:
1.68 kristaps 733: abort();
734: /* NOTREACHED */
1.1 kristaps 735: }
736:
737: return(1);
738: }
CVSweb