Annotation of mandoc/mdoc_argv.c, Revision 1.65
1.65 ! kristaps 1: /* $Id: mdoc_argv.c,v 1.64 2011/03/17 00:58:14 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:
1.49 kristaps 29: #include "mandoc.h"
1.1 kristaps 30: #include "libmdoc.h"
1.31 kristaps 31: #include "libmandoc.h"
1.1 kristaps 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:
1.57 kristaps 48: static enum mdocargt argv_a2arg(enum mdoct, const char *);
1.39 kristaps 49: static enum margserr args(struct mdoc *, int, int *,
1.1 kristaps 50: char *, int, char **);
1.64 kristaps 51: static int args_checkpunct(const char *);
1.1 kristaps 52: static int argv(struct mdoc *, int,
53: struct mdoc_argv *, int *, char *);
54: static int argv_single(struct mdoc *, int,
55: struct mdoc_argv *, int *, char *);
56: static int argv_opt_single(struct mdoc *, int,
57: struct mdoc_argv *, int *, char *);
58: static int argv_multi(struct mdoc *, int,
59: struct mdoc_argv *, int *, char *);
1.2 kristaps 60:
1.1 kristaps 61: /* Per-argument flags. */
62:
63: static int mdoc_argvflags[MDOC_ARG_MAX] = {
64: ARGV_NONE, /* MDOC_Split */
65: ARGV_NONE, /* MDOC_Nosplit */
66: ARGV_NONE, /* MDOC_Ragged */
67: ARGV_NONE, /* MDOC_Unfilled */
68: ARGV_NONE, /* MDOC_Literal */
1.29 kristaps 69: ARGV_SINGLE, /* MDOC_File */
1.26 kristaps 70: ARGV_OPT_SINGLE, /* MDOC_Offset */
1.1 kristaps 71: ARGV_NONE, /* MDOC_Bullet */
72: ARGV_NONE, /* MDOC_Dash */
73: ARGV_NONE, /* MDOC_Hyphen */
74: ARGV_NONE, /* MDOC_Item */
75: ARGV_NONE, /* MDOC_Enum */
76: ARGV_NONE, /* MDOC_Tag */
77: ARGV_NONE, /* MDOC_Diag */
78: ARGV_NONE, /* MDOC_Hang */
79: ARGV_NONE, /* MDOC_Ohang */
80: ARGV_NONE, /* MDOC_Inset */
81: ARGV_MULTI, /* MDOC_Column */
82: ARGV_SINGLE, /* MDOC_Width */
83: ARGV_NONE, /* MDOC_Compact */
1.25 kristaps 84: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 85: ARGV_NONE, /* MDOC_Filled */
86: ARGV_NONE, /* MDOC_Words */
87: ARGV_NONE, /* MDOC_Emphasis */
88: ARGV_NONE, /* MDOC_Symbolic */
89: ARGV_NONE /* MDOC_Symbolic */
90: };
91:
92: static int mdoc_argflags[MDOC_MAX] = {
1.5 kristaps 93: 0, /* Ap */
1.1 kristaps 94: 0, /* Dd */
95: 0, /* Dt */
96: 0, /* Os */
1.17 kristaps 97: 0, /* Sh */
98: 0, /* Ss */
1.59 kristaps 99: 0, /* Pp */
1.1 kristaps 100: ARGS_DELIM, /* D1 */
1.17 kristaps 101: ARGS_DELIM, /* Dl */
1.1 kristaps 102: 0, /* Bd */
103: 0, /* Ed */
1.17 kristaps 104: 0, /* Bl */
1.1 kristaps 105: 0, /* El */
106: 0, /* It */
107: ARGS_DELIM, /* Ad */
108: ARGS_DELIM, /* An */
1.17 kristaps 109: ARGS_DELIM, /* Ar */
110: 0, /* Cd */
1.1 kristaps 111: ARGS_DELIM, /* Cm */
112: ARGS_DELIM, /* Dv */
113: ARGS_DELIM, /* Er */
114: ARGS_DELIM, /* Ev */
115: 0, /* Ex */
1.17 kristaps 116: ARGS_DELIM, /* Fa */
1.1 kristaps 117: 0, /* Fd */
118: ARGS_DELIM, /* Fl */
1.17 kristaps 119: ARGS_DELIM, /* Fn */
120: ARGS_DELIM, /* Ft */
1.1 kristaps 121: ARGS_DELIM, /* Ic */
122: 0, /* In */
1.17 kristaps 123: ARGS_DELIM, /* Li */
124: 0, /* Nd */
1.1 kristaps 125: ARGS_DELIM, /* Nm */
126: ARGS_DELIM, /* Op */
127: 0, /* Ot */
128: ARGS_DELIM, /* Pa */
129: 0, /* Rv */
1.15 kristaps 130: ARGS_DELIM, /* St */
1.1 kristaps 131: ARGS_DELIM, /* Va */
132: ARGS_DELIM, /* Vt */
133: ARGS_DELIM, /* Xr */
1.17 kristaps 134: 0, /* %A */
135: 0, /* %B */
136: 0, /* %D */
137: 0, /* %I */
138: 0, /* %J */
139: 0, /* %N */
140: 0, /* %O */
141: 0, /* %P */
142: 0, /* %R */
143: 0, /* %T */
144: 0, /* %V */
1.1 kristaps 145: ARGS_DELIM, /* Ac */
146: 0, /* Ao */
147: ARGS_DELIM, /* Aq */
148: ARGS_DELIM, /* At */
149: ARGS_DELIM, /* Bc */
150: 0, /* Bf */
151: 0, /* Bo */
152: ARGS_DELIM, /* Bq */
153: ARGS_DELIM, /* Bsx */
154: ARGS_DELIM, /* Bx */
155: 0, /* Db */
156: ARGS_DELIM, /* Dc */
157: 0, /* Do */
158: ARGS_DELIM, /* Dq */
159: ARGS_DELIM, /* Ec */
160: 0, /* Ef */
161: ARGS_DELIM, /* Em */
162: 0, /* Eo */
163: ARGS_DELIM, /* Fx */
164: ARGS_DELIM, /* Ms */
165: ARGS_DELIM, /* No */
166: ARGS_DELIM, /* Ns */
167: ARGS_DELIM, /* Nx */
168: ARGS_DELIM, /* Ox */
169: ARGS_DELIM, /* Pc */
170: ARGS_DELIM, /* Pf */
171: 0, /* Po */
172: ARGS_DELIM, /* Pq */
173: ARGS_DELIM, /* Qc */
174: ARGS_DELIM, /* Ql */
175: 0, /* Qo */
176: ARGS_DELIM, /* Qq */
177: 0, /* Re */
178: 0, /* Rs */
179: ARGS_DELIM, /* Sc */
180: 0, /* So */
181: ARGS_DELIM, /* Sq */
182: 0, /* Sm */
183: ARGS_DELIM, /* Sx */
1.17 kristaps 184: ARGS_DELIM, /* Sy */
1.1 kristaps 185: ARGS_DELIM, /* Tn */
186: ARGS_DELIM, /* Ux */
187: ARGS_DELIM, /* Xc */
188: 0, /* Xo */
1.17 kristaps 189: 0, /* Fo */
1.1 kristaps 190: 0, /* Fc */
191: 0, /* Oo */
192: ARGS_DELIM, /* Oc */
193: 0, /* Bk */
194: 0, /* Ek */
195: 0, /* Bt */
196: 0, /* Hf */
197: 0, /* Fr */
198: 0, /* Ud */
199: 0, /* Lb */
1.60 schwarze 200: 0, /* Lp */
1.17 kristaps 201: ARGS_DELIM, /* Lk */
202: ARGS_DELIM, /* Mt */
1.1 kristaps 203: ARGS_DELIM, /* Brq */
204: 0, /* Bro */
205: ARGS_DELIM, /* Brc */
1.17 kristaps 206: 0, /* %C */
1.1 kristaps 207: 0, /* Es */
208: 0, /* En */
209: 0, /* Dx */
1.17 kristaps 210: 0, /* %Q */
1.14 kristaps 211: 0, /* br */
212: 0, /* sp */
1.30 kristaps 213: 0, /* %U */
1.51 kristaps 214: 0, /* Ta */
1.1 kristaps 215: };
216:
217:
218: /*
219: * Parse an argument from line text. This comes in the form of -key
220: * [value0...], which may either have a single mandatory value, at least
221: * one mandatory value, an optional single value, or no value.
222: */
1.40 kristaps 223: enum margverr
1.35 kristaps 224: mdoc_argv(struct mdoc *m, int line, enum mdoct tok,
1.1 kristaps 225: struct mdoc_arg **v, int *pos, char *buf)
226: {
227: char *p, sv;
228: struct mdoc_argv tmp;
229: struct mdoc_arg *arg;
230:
1.52 kristaps 231: if ('\0' == buf[*pos])
1.1 kristaps 232: return(ARGV_EOLN);
233:
234: assert(' ' != buf[*pos]);
235:
236: /* Parse through to the first unescaped space. */
237:
238: p = &buf[++(*pos)];
239:
240: assert(*pos > 0);
241:
242: /* LINTED */
243: while (buf[*pos]) {
244: if (' ' == buf[*pos])
245: if ('\\' != buf[*pos - 1])
246: break;
247: (*pos)++;
248: }
249:
250: /* XXX - save zeroed byte, if not an argument. */
251:
1.52 kristaps 252: sv = '\0';
1.1 kristaps 253: if (buf[*pos]) {
254: sv = buf[*pos];
1.52 kristaps 255: buf[(*pos)++] = '\0';
1.1 kristaps 256: }
257:
258: (void)memset(&tmp, 0, sizeof(struct mdoc_argv));
259: tmp.line = line;
260: tmp.pos = *pos;
261:
262: /* See if our token accepts the argument. */
263:
264: if (MDOC_ARG_MAX == (tmp.arg = argv_a2arg(tok, p))) {
265: /* XXX - restore saved zeroed byte. */
266: if (sv)
267: buf[*pos - 1] = sv;
268: return(ARGV_WORD);
269: }
270:
271: while (buf[*pos] && ' ' == buf[*pos])
272: (*pos)++;
273:
1.11 kristaps 274: if ( ! argv(m, line, &tmp, pos, buf))
1.1 kristaps 275: return(ARGV_ERROR);
276:
1.31 kristaps 277: if (NULL == (arg = *v))
1.32 kristaps 278: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 279:
280: arg->argc++;
1.31 kristaps 281: arg->argv = mandoc_realloc
282: (arg->argv, arg->argc * sizeof(struct mdoc_argv));
1.2 kristaps 283:
1.1 kristaps 284: (void)memcpy(&arg->argv[(int)arg->argc - 1],
285: &tmp, sizeof(struct mdoc_argv));
286:
287: return(ARGV_ARG);
288: }
289:
290:
291: void
292: mdoc_argv_free(struct mdoc_arg *p)
293: {
1.37 kristaps 294: int i;
1.1 kristaps 295:
296: if (NULL == p)
297: return;
298:
299: if (p->refcnt) {
300: --(p->refcnt);
301: if (p->refcnt)
302: return;
303: }
304: assert(p->argc);
305:
1.37 kristaps 306: for (i = (int)p->argc - 1; i >= 0; i--)
307: mdoc_argn_free(p, i);
308:
309: free(p->argv);
310: free(p);
311: }
312:
313:
314: void
315: mdoc_argn_free(struct mdoc_arg *p, int iarg)
316: {
1.58 kristaps 317: struct mdoc_argv *arg;
1.37 kristaps 318: int j;
1.58 kristaps 319:
320: arg = &p->argv[iarg];
1.1 kristaps 321:
1.37 kristaps 322: if (arg->sz && arg->value) {
323: for (j = (int)arg->sz - 1; j >= 0; j--)
324: free(arg->value[j]);
325: free(arg->value);
1.1 kristaps 326: }
327:
1.37 kristaps 328: for (--p->argc; iarg < (int)p->argc; iarg++)
329: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 330: }
331:
332:
1.39 kristaps 333: enum margserr
1.23 kristaps 334: mdoc_zargs(struct mdoc *m, int line, int *pos,
335: char *buf, int flags, char **v)
1.19 kristaps 336: {
337:
1.23 kristaps 338: return(args(m, line, pos, buf, flags, v));
1.19 kristaps 339: }
340:
341:
1.39 kristaps 342: enum margserr
1.35 kristaps 343: mdoc_args(struct mdoc *m, int line, int *pos,
344: char *buf, enum mdoct tok, char **v)
1.1 kristaps 345: {
1.50 kristaps 346: int fl;
1.1 kristaps 347: struct mdoc_node *n;
348:
1.35 kristaps 349: fl = mdoc_argflags[tok];
1.1 kristaps 350:
1.17 kristaps 351: if (MDOC_It != tok)
352: return(args(m, line, pos, buf, fl, v));
353:
1.51 kristaps 354: /*
355: * We know that we're in an `It', so it's reasonable to expect
356: * us to be sitting in a `Bl'. Someday this may not be the case
357: * (if we allow random `It's sitting out there), so provide a
358: * safe fall-back into the default behaviour.
359: */
360:
1.17 kristaps 361: for (n = m->last; n; n = n->parent)
1.50 kristaps 362: if (MDOC_Bl == n->tok)
1.17 kristaps 363: break;
1.1 kristaps 364:
1.62 kristaps 365: if (n && LIST_column == n->norm->Bl.type) {
1.17 kristaps 366: fl |= ARGS_TABSEP;
367: fl &= ~ARGS_DELIM;
1.1 kristaps 368: }
369:
1.11 kristaps 370: return(args(m, line, pos, buf, fl, v));
1.1 kristaps 371: }
372:
373:
1.39 kristaps 374: static enum margserr
1.13 kristaps 375: args(struct mdoc *m, int line, int *pos,
376: char *buf, int fl, char **v)
1.1 kristaps 377: {
1.52 kristaps 378: int i;
379: char *p, *pp;
380: enum margserr rc;
1.1 kristaps 381:
1.24 kristaps 382: /*
383: * Parse out the terms (like `val' in `.Xx -arg val' or simply
384: * `.Xx val'), which can have all sorts of properties:
385: *
386: * ARGS_DELIM: use special handling if encountering trailing
387: * delimiters in the form of [[::delim::][ ]+]+.
388: *
389: * ARGS_NOWARN: don't post warnings. This is only used when
390: * re-parsing delimiters, as the warnings have already been
391: * posted.
392: *
393: * ARGS_TABSEP: use special handling for tab/`Ta' separated
394: * phrases like in `Bl -column'.
395: */
396:
1.17 kristaps 397: assert(' ' != buf[*pos]);
1.1 kristaps 398:
1.46 kristaps 399: if ('\0' == buf[*pos]) {
1.47 kristaps 400: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 401: return(ARGS_EOLN);
402: /*
403: * If we're not in a partial phrase and the flag for
404: * being a phrase literal is still set, the punctuation
405: * is unterminated.
406: */
407: if (MDOC_PHRASELIT & m->flags)
1.49 kristaps 408: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE))
1.46 kristaps 409: return(ARGS_ERROR);
410:
411: m->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 412: return(ARGS_EOLN);
1.46 kristaps 413: }
1.1 kristaps 414:
1.64 kristaps 415: *v = &buf[*pos];
1.17 kristaps 416:
1.64 kristaps 417: if (ARGS_DELIM & fl && args_checkpunct(&buf[*pos])) {
418: i = strlen(&buf[*pos]) + *pos;
419: if (i && ' ' != buf[i - 1])
420: return(ARGS_PUNCT);
421: if (ARGS_NOWARN & fl)
1.1 kristaps 422: return(ARGS_PUNCT);
1.64 kristaps 423: /* FIXME: remove conditional messages... */
424: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
425: return(ARGS_ERROR);
426: return(ARGS_PUNCT);
1.1 kristaps 427: }
428:
1.17 kristaps 429: /*
430: * First handle TABSEP items, restricted to `Bl -column'. This
431: * ignores conventional token parsing and instead uses tabs or
432: * `Ta' macros to separate phrases. Phrases are parsed again
433: * for arguments at a later phase.
434: */
1.1 kristaps 435:
1.17 kristaps 436: if (ARGS_TABSEP & fl) {
1.20 kristaps 437: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 438: p = strchr(*v, '\t');
1.46 kristaps 439: pp = NULL;
1.17 kristaps 440:
441: /* Scan ahead to unescaped `Ta'. */
1.46 kristaps 442: if ( ! (MDOC_PHRASELIT & m->flags))
443: for (pp = *v; ; pp++) {
444: if (NULL == (pp = strstr(pp, "Ta")))
445: break;
446: if (pp > *v && ' ' != *(pp - 1))
447: continue;
1.51 kristaps 448: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 449: break;
450: }
1.17 kristaps 451:
1.42 kristaps 452: /* By default, assume a phrase. */
453: rc = ARGS_PHRASE;
454:
1.21 kristaps 455: /*
456: * Adjust new-buffer position to be beyond delimiter
457: * mark (e.g., Ta -> end + 2).
458: */
1.20 kristaps 459: if (p && pp) {
460: *pos += pp < p ? 2 : 1;
1.42 kristaps 461: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 462: p = pp < p ? pp : p;
463: } else if (p && ! pp) {
1.42 kristaps 464: rc = ARGS_PPHRASE;
1.20 kristaps 465: *pos += 1;
466: } else if (pp && ! p) {
1.17 kristaps 467: p = pp;
1.20 kristaps 468: *pos += 2;
1.44 kristaps 469: } else {
470: rc = ARGS_PEND;
1.20 kristaps 471: p = strchr(*v, 0);
1.44 kristaps 472: }
1.20 kristaps 473:
1.21 kristaps 474: /* Whitespace check for eoln case... */
1.52 kristaps 475: if ('\0' == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl))
1.49 kristaps 476: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
1.20 kristaps 477: return(ARGS_ERROR);
478:
479: *pos += (int)(p - *v);
1.17 kristaps 480:
481: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 482: pp = p - 1;
483: while (pp > *v && ' ' == *pp) {
484: if (pp > *v && '\\' == *(pp - 1))
485: break;
486: pp--;
1.17 kristaps 487: }
1.20 kristaps 488: *(pp + 1) = 0;
1.17 kristaps 489:
1.20 kristaps 490: /* Strip delimiter's proceeding whitespace. */
491: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
492: /* Skip ahead. */ ;
1.1 kristaps 493:
1.41 kristaps 494: return(rc);
1.17 kristaps 495: }
1.1 kristaps 496:
1.17 kristaps 497: /*
498: * Process a quoted literal. A quote begins with a double-quote
499: * and ends with a double-quote NOT preceded by a double-quote.
500: * Whitespace is NOT involved in literal termination.
501: */
1.1 kristaps 502:
1.46 kristaps 503: if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
504: if ( ! (MDOC_PHRASELIT & m->flags))
505: *v = &buf[++(*pos)];
506:
1.47 kristaps 507: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 508: m->flags |= MDOC_PHRASELIT;
1.1 kristaps 509:
1.17 kristaps 510: for ( ; buf[*pos]; (*pos)++) {
511: if ('\"' != buf[*pos])
512: continue;
513: if ('\"' != buf[*pos + 1])
514: break;
515: (*pos)++;
516: }
1.1 kristaps 517:
1.46 kristaps 518: if ('\0' == buf[*pos]) {
1.47 kristaps 519: if (ARGS_NOWARN & fl || MDOC_PPHRASE & m->flags)
1.23 kristaps 520: return(ARGS_QWORD);
1.49 kristaps 521: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE))
1.18 kristaps 522: return(ARGS_ERROR);
523: return(ARGS_QWORD);
1.17 kristaps 524: }
1.1 kristaps 525:
1.46 kristaps 526: m->flags &= ~MDOC_PHRASELIT;
527: buf[(*pos)++] = '\0';
1.1 kristaps 528:
1.46 kristaps 529: if ('\0' == buf[*pos])
1.17 kristaps 530: return(ARGS_QWORD);
1.1 kristaps 531:
1.17 kristaps 532: while (' ' == buf[*pos])
533: (*pos)++;
1.1 kristaps 534:
1.23 kristaps 535: if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
1.49 kristaps 536: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
1.17 kristaps 537: return(ARGS_ERROR);
1.1 kristaps 538:
1.17 kristaps 539: return(ARGS_QWORD);
1.1 kristaps 540: }
541:
1.17 kristaps 542: /*
543: * A non-quoted term progresses until either the end of line or
544: * a non-escaped whitespace.
1.1 kristaps 545: */
546:
1.17 kristaps 547: for ( ; buf[*pos]; (*pos)++)
1.52 kristaps 548: if (*pos && ' ' == buf[*pos] && '\\' != buf[*pos - 1])
1.17 kristaps 549: break;
1.1 kristaps 550:
1.46 kristaps 551: if ('\0' == buf[*pos])
1.17 kristaps 552: return(ARGS_WORD);
1.1 kristaps 553:
1.46 kristaps 554: buf[(*pos)++] = '\0';
1.1 kristaps 555:
1.17 kristaps 556: while (' ' == buf[*pos])
1.1 kristaps 557: (*pos)++;
558:
1.46 kristaps 559: if ('\0' == buf[*pos] && ! (ARGS_NOWARN & fl))
1.49 kristaps 560: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
1.17 kristaps 561: return(ARGS_ERROR);
1.1 kristaps 562:
1.17 kristaps 563: return(ARGS_WORD);
1.64 kristaps 564: }
565:
566: /*
567: * Check if the string consists only of space-separated closing
1.65 ! kristaps 568: * delimiters. This is a bit of a dance: the first must be a close
! 569: * delimiter, but it may be followed by middle delimiters. Arbitrary
! 570: * whitespace may separate these tokens.
1.64 kristaps 571: */
572: static int
573: args_checkpunct(const char *p)
574: {
1.65 ! kristaps 575: int i, j;
! 576: char buf[DELIMSZ];
1.64 kristaps 577: enum mdelim d;
578:
579: i = 0;
580:
1.65 ! kristaps 581: /* First token must be a close-delimiter. */
! 582:
! 583: for (j = 0; p[i] && ' ' != p[i] && j < DELIMSZ; j++, i++)
! 584: buf[j] = p[i];
! 585:
! 586: if (DELIMSZ == j)
! 587: return(0);
! 588:
! 589: buf[j] = '\0';
! 590: if (DELIM_CLOSE != mdoc_isdelim(buf))
1.64 kristaps 591: return(0);
592:
1.65 ! kristaps 593: while (' ' == p[i])
! 594: i++;
! 595:
! 596: /* Remaining must NOT be open/none. */
! 597:
! 598: while (p[i]) {
! 599: j = 0;
! 600: while (p[i] && ' ' != p[i] && j < DELIMSZ)
! 601: buf[j++] = p[i++];
! 602:
! 603: if (DELIMSZ == j)
! 604: return(0);
! 605:
! 606: buf[j] = '\0';
! 607: d = mdoc_isdelim(buf);
1.64 kristaps 608: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 ! kristaps 609: return(0);
! 610:
! 611: while (' ' == p[i])
1.64 kristaps 612: i++;
613: }
614:
615: return('\0' == p[i]);
1.1 kristaps 616: }
617:
1.63 kristaps 618: /*
619: * Match up an argument string (e.g., `-foo bar' having "foo") with the
620: * correrct identifier. It must apply to the given macro. If none was
621: * found (including bad matches), return MDOC_ARG_MAX.
622: */
1.57 kristaps 623: static enum mdocargt
1.35 kristaps 624: argv_a2arg(enum mdoct tok, const char *p)
1.1 kristaps 625: {
1.63 kristaps 626: enum mdocargt args[MDOC_ARG_MAX];
627: int i, len;
1.1 kristaps 628:
1.63 kristaps 629: len = 0;
1.1 kristaps 630:
631: switch (tok) {
632: case (MDOC_An):
1.63 kristaps 633: args[len++] = MDOC_Split;
634: args[len++] = MDOC_Nosplit;
1.1 kristaps 635: break;
636: case (MDOC_Bd):
1.63 kristaps 637: args[len++] = MDOC_Ragged;
638: args[len++] = MDOC_Unfilled;
639: args[len++] = MDOC_Filled;
640: args[len++] = MDOC_Literal;
641: args[len++] = MDOC_File;
642: args[len++] = MDOC_Offset;
643: args[len++] = MDOC_Compact;
644: args[len++] = MDOC_Centred;
1.1 kristaps 645: break;
646: case (MDOC_Bf):
1.63 kristaps 647: args[len++] = MDOC_Emphasis;
648: args[len++] = MDOC_Literal;
649: args[len++] = MDOC_Symbolic;
1.1 kristaps 650: break;
651: case (MDOC_Bk):
1.63 kristaps 652: args[len++] = MDOC_Words;
1.1 kristaps 653: break;
654: case (MDOC_Bl):
1.63 kristaps 655: args[len++] = MDOC_Bullet;
656: args[len++] = MDOC_Dash;
657: args[len++] = MDOC_Hyphen;
658: args[len++] = MDOC_Item;
659: args[len++] = MDOC_Enum;
660: args[len++] = MDOC_Tag;
661: args[len++] = MDOC_Diag;
662: args[len++] = MDOC_Hang;
663: args[len++] = MDOC_Ohang;
664: args[len++] = MDOC_Inset;
665: args[len++] = MDOC_Column;
666: args[len++] = MDOC_Width;
667: args[len++] = MDOC_Offset;
668: args[len++] = MDOC_Compact;
669: args[len++] = MDOC_Nested;
1.1 kristaps 670: break;
671: case (MDOC_Rv):
672: /* FALLTHROUGH */
673: case (MDOC_Ex):
1.63 kristaps 674: args[len++] = MDOC_Std;
1.1 kristaps 675: break;
676: default:
677: break;
678: }
1.63 kristaps 679:
680: for (i = 0; i < len; i++)
681: if (0 == strcmp(p, mdoc_argnames[args[i]]))
682: return(args[i]);
1.1 kristaps 683:
684: return(MDOC_ARG_MAX);
685: }
686:
687:
688: static int
1.11 kristaps 689: argv_multi(struct mdoc *m, int line,
1.1 kristaps 690: struct mdoc_argv *v, int *pos, char *buf)
691: {
1.43 kristaps 692: enum margserr ac;
1.1 kristaps 693: char *p;
694:
695: for (v->sz = 0; ; v->sz++) {
696: if ('-' == buf[*pos])
697: break;
1.43 kristaps 698: ac = args(m, line, pos, buf, 0, &p);
699: if (ARGS_ERROR == ac)
1.1 kristaps 700: return(0);
1.43 kristaps 701: else if (ARGS_EOLN == ac)
1.1 kristaps 702: break;
703:
1.31 kristaps 704: if (0 == v->sz % MULTI_STEP)
705: v->value = mandoc_realloc(v->value,
1.1 kristaps 706: (v->sz + MULTI_STEP) * sizeof(char *));
1.31 kristaps 707:
708: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 709: }
710:
1.7 kristaps 711: return(1);
1.1 kristaps 712: }
713:
714:
715: static int
1.11 kristaps 716: argv_opt_single(struct mdoc *m, int line,
1.1 kristaps 717: struct mdoc_argv *v, int *pos, char *buf)
718: {
1.43 kristaps 719: enum margserr ac;
1.1 kristaps 720: char *p;
721:
722: if ('-' == buf[*pos])
723: return(1);
724:
1.43 kristaps 725: ac = args(m, line, pos, buf, 0, &p);
726: if (ARGS_ERROR == ac)
1.1 kristaps 727: return(0);
1.43 kristaps 728: if (ARGS_EOLN == ac)
1.1 kristaps 729: return(1);
730:
731: v->sz = 1;
1.31 kristaps 732: v->value = mandoc_malloc(sizeof(char *));
733: v->value[0] = mandoc_strdup(p);
1.2 kristaps 734:
1.1 kristaps 735: return(1);
736: }
737:
738:
739: /*
740: * Parse a single, mandatory value from the stream.
741: */
742: static int
1.11 kristaps 743: argv_single(struct mdoc *m, int line,
1.1 kristaps 744: struct mdoc_argv *v, int *pos, char *buf)
745: {
1.43 kristaps 746: int ppos;
747: enum margserr ac;
1.1 kristaps 748: char *p;
749:
750: ppos = *pos;
751:
1.43 kristaps 752: ac = args(m, line, pos, buf, 0, &p);
1.49 kristaps 753: if (ARGS_EOLN == ac) {
754: mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTARGVCOUNT);
755: return(0);
756: } else if (ARGS_ERROR == ac)
1.1 kristaps 757: return(0);
758:
759: v->sz = 1;
1.31 kristaps 760: v->value = mandoc_malloc(sizeof(char *));
761: v->value[0] = mandoc_strdup(p);
1.2 kristaps 762:
1.1 kristaps 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: }
CVSweb