Annotation of mandoc/mdoc_argv.c, Revision 1.55
1.55 ! kristaps 1: /* $Id: mdoc_argv.c,v 1.54 2010/06/19 20:46:28 kristaps Exp $ */
1.1 kristaps 2: /*
1.54 kristaps 3: * Copyright (c) 2008, 2009 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:
1.35 kristaps 48: static int 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 **);
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 *);
1.2 kristaps 59:
1.1 kristaps 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 */
1.29 kristaps 68: ARGV_SINGLE, /* MDOC_File */
1.26 kristaps 69: ARGV_OPT_SINGLE, /* MDOC_Offset */
1.1 kristaps 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 */
1.25 kristaps 83: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 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] = {
1.5 kristaps 92: 0, /* Ap */
1.1 kristaps 93: 0, /* Dd */
94: 0, /* Dt */
95: 0, /* Os */
1.17 kristaps 96: 0, /* Sh */
97: 0, /* Ss */
1.1 kristaps 98: ARGS_DELIM, /* Pp */
99: ARGS_DELIM, /* D1 */
1.17 kristaps 100: ARGS_DELIM, /* Dl */
1.1 kristaps 101: 0, /* Bd */
102: 0, /* Ed */
1.17 kristaps 103: 0, /* Bl */
1.1 kristaps 104: 0, /* El */
105: 0, /* It */
106: ARGS_DELIM, /* Ad */
107: ARGS_DELIM, /* An */
1.17 kristaps 108: ARGS_DELIM, /* Ar */
109: 0, /* Cd */
1.1 kristaps 110: ARGS_DELIM, /* Cm */
111: ARGS_DELIM, /* Dv */
112: ARGS_DELIM, /* Er */
113: ARGS_DELIM, /* Ev */
114: 0, /* Ex */
1.17 kristaps 115: ARGS_DELIM, /* Fa */
1.1 kristaps 116: 0, /* Fd */
117: ARGS_DELIM, /* Fl */
1.17 kristaps 118: ARGS_DELIM, /* Fn */
119: ARGS_DELIM, /* Ft */
1.1 kristaps 120: ARGS_DELIM, /* Ic */
121: 0, /* In */
1.17 kristaps 122: ARGS_DELIM, /* Li */
123: 0, /* Nd */
1.1 kristaps 124: ARGS_DELIM, /* Nm */
125: ARGS_DELIM, /* Op */
126: 0, /* Ot */
127: ARGS_DELIM, /* Pa */
128: 0, /* Rv */
1.15 kristaps 129: ARGS_DELIM, /* St */
1.1 kristaps 130: ARGS_DELIM, /* Va */
131: ARGS_DELIM, /* Vt */
132: ARGS_DELIM, /* Xr */
1.17 kristaps 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 */
1.1 kristaps 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 */
1.17 kristaps 183: ARGS_DELIM, /* Sy */
1.1 kristaps 184: ARGS_DELIM, /* Tn */
185: ARGS_DELIM, /* Ux */
186: ARGS_DELIM, /* Xc */
187: 0, /* Xo */
1.17 kristaps 188: 0, /* Fo */
1.1 kristaps 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: ARGS_DELIM, /* Lp */
1.17 kristaps 200: ARGS_DELIM, /* Lk */
201: ARGS_DELIM, /* Mt */
1.1 kristaps 202: ARGS_DELIM, /* Brq */
203: 0, /* Bro */
204: ARGS_DELIM, /* Brc */
1.17 kristaps 205: 0, /* %C */
1.1 kristaps 206: 0, /* Es */
207: 0, /* En */
208: 0, /* Dx */
1.17 kristaps 209: 0, /* %Q */
1.14 kristaps 210: 0, /* br */
211: 0, /* sp */
1.30 kristaps 212: 0, /* %U */
1.51 kristaps 213: 0, /* Ta */
1.1 kristaps 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: */
1.40 kristaps 222: enum margverr
1.35 kristaps 223: mdoc_argv(struct mdoc *m, int line, enum mdoct tok,
1.1 kristaps 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:
1.52 kristaps 230: if ('\0' == buf[*pos])
1.1 kristaps 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:
1.52 kristaps 251: sv = '\0';
1.1 kristaps 252: if (buf[*pos]) {
253: sv = buf[*pos];
1.52 kristaps 254: buf[(*pos)++] = '\0';
1.1 kristaps 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:
1.11 kristaps 273: if ( ! argv(m, line, &tmp, pos, buf))
1.1 kristaps 274: return(ARGV_ERROR);
275:
1.31 kristaps 276: if (NULL == (arg = *v))
1.32 kristaps 277: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 278:
279: arg->argc++;
1.31 kristaps 280: arg->argv = mandoc_realloc
281: (arg->argv, arg->argc * sizeof(struct mdoc_argv));
1.2 kristaps 282:
1.1 kristaps 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: {
1.37 kristaps 293: int i;
1.1 kristaps 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:
1.37 kristaps 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 = &p->argv[iarg];
317: int j;
1.1 kristaps 318:
1.37 kristaps 319: if (arg->sz && arg->value) {
320: for (j = (int)arg->sz - 1; j >= 0; j--)
321: free(arg->value[j]);
322: free(arg->value);
1.1 kristaps 323: }
324:
1.37 kristaps 325: for (--p->argc; iarg < (int)p->argc; iarg++)
326: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 327: }
328:
329:
1.39 kristaps 330: enum margserr
1.23 kristaps 331: mdoc_zargs(struct mdoc *m, int line, int *pos,
332: char *buf, int flags, char **v)
1.19 kristaps 333: {
334:
1.23 kristaps 335: return(args(m, line, pos, buf, flags, v));
1.19 kristaps 336: }
337:
338:
1.39 kristaps 339: enum margserr
1.35 kristaps 340: mdoc_args(struct mdoc *m, int line, int *pos,
341: char *buf, enum mdoct tok, char **v)
1.1 kristaps 342: {
1.50 kristaps 343: int fl;
1.1 kristaps 344: struct mdoc_node *n;
345:
1.35 kristaps 346: fl = mdoc_argflags[tok];
1.1 kristaps 347:
1.17 kristaps 348: if (MDOC_It != tok)
349: return(args(m, line, pos, buf, fl, v));
350:
1.51 kristaps 351: /*
352: * We know that we're in an `It', so it's reasonable to expect
353: * us to be sitting in a `Bl'. Someday this may not be the case
354: * (if we allow random `It's sitting out there), so provide a
355: * safe fall-back into the default behaviour.
356: */
357:
1.17 kristaps 358: for (n = m->last; n; n = n->parent)
1.50 kristaps 359: if (MDOC_Bl == n->tok)
1.17 kristaps 360: break;
1.1 kristaps 361:
1.55 ! kristaps 362: assert(n->data.Bl);
! 363: if (n && LIST_column == n->data.Bl->type) {
1.17 kristaps 364: fl |= ARGS_TABSEP;
365: fl &= ~ARGS_DELIM;
1.1 kristaps 366: }
367:
1.11 kristaps 368: return(args(m, line, pos, buf, fl, v));
1.1 kristaps 369: }
370:
371:
1.39 kristaps 372: static enum margserr
1.13 kristaps 373: args(struct mdoc *m, int line, int *pos,
374: char *buf, int fl, char **v)
1.1 kristaps 375: {
1.52 kristaps 376: int i;
377: char *p, *pp;
378: enum margserr rc;
379: enum mdelim d;
1.1 kristaps 380:
1.24 kristaps 381: /*
382: * Parse out the terms (like `val' in `.Xx -arg val' or simply
383: * `.Xx val'), which can have all sorts of properties:
384: *
385: * ARGS_DELIM: use special handling if encountering trailing
386: * delimiters in the form of [[::delim::][ ]+]+.
387: *
388: * ARGS_NOWARN: don't post warnings. This is only used when
389: * re-parsing delimiters, as the warnings have already been
390: * posted.
391: *
392: * ARGS_TABSEP: use special handling for tab/`Ta' separated
393: * phrases like in `Bl -column'.
394: */
395:
1.17 kristaps 396: assert(' ' != buf[*pos]);
1.1 kristaps 397:
1.46 kristaps 398: if ('\0' == buf[*pos]) {
1.47 kristaps 399: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 400: return(ARGS_EOLN);
401: /*
402: * If we're not in a partial phrase and the flag for
403: * being a phrase literal is still set, the punctuation
404: * is unterminated.
405: */
406: if (MDOC_PHRASELIT & m->flags)
1.49 kristaps 407: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE))
1.46 kristaps 408: return(ARGS_ERROR);
409:
410: m->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 411: return(ARGS_EOLN);
1.46 kristaps 412: }
1.1 kristaps 413:
414: /*
1.45 kristaps 415: * If the first character is a closing delimiter and we're to
416: * look for delimited strings, then pass down the buffer seeing
417: * if it follows the pattern of [[::delim::][ ]+]+. Note that
418: * we ONLY care about closing delimiters.
1.1 kristaps 419: */
420:
1.48 schwarze 421: if ((fl & ARGS_DELIM) && DELIM_CLOSE == mdoc_iscdelim(buf[*pos])) {
1.1 kristaps 422: for (i = *pos; buf[i]; ) {
1.52 kristaps 423: d = mdoc_iscdelim(buf[i]);
1.48 schwarze 424: if (DELIM_NONE == d || DELIM_OPEN == d)
1.1 kristaps 425: break;
426: i++;
1.45 kristaps 427: if ('\0' == buf[i] || ' ' != buf[i])
1.1 kristaps 428: break;
429: i++;
430: while (buf[i] && ' ' == buf[i])
431: i++;
432: }
1.17 kristaps 433:
1.45 kristaps 434: if ('\0' == buf[i]) {
1.1 kristaps 435: *v = &buf[*pos];
1.52 kristaps 436: if (i && ' ' != buf[i - 1])
1.23 kristaps 437: return(ARGS_PUNCT);
438: if (ARGS_NOWARN & fl)
439: return(ARGS_PUNCT);
1.49 kristaps 440: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
1.23 kristaps 441: return(ARGS_ERROR);
1.1 kristaps 442: return(ARGS_PUNCT);
443: }
444: }
445:
1.17 kristaps 446: *v = &buf[*pos];
1.1 kristaps 447:
1.17 kristaps 448: /*
449: * First handle TABSEP items, restricted to `Bl -column'. This
450: * ignores conventional token parsing and instead uses tabs or
451: * `Ta' macros to separate phrases. Phrases are parsed again
452: * for arguments at a later phase.
453: */
1.1 kristaps 454:
1.17 kristaps 455: if (ARGS_TABSEP & fl) {
1.20 kristaps 456: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 457: p = strchr(*v, '\t');
1.46 kristaps 458: pp = NULL;
1.17 kristaps 459:
460: /* Scan ahead to unescaped `Ta'. */
1.46 kristaps 461: if ( ! (MDOC_PHRASELIT & m->flags))
462: for (pp = *v; ; pp++) {
463: if (NULL == (pp = strstr(pp, "Ta")))
464: break;
465: if (pp > *v && ' ' != *(pp - 1))
466: continue;
1.51 kristaps 467: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 468: break;
469: }
1.17 kristaps 470:
1.42 kristaps 471: /* By default, assume a phrase. */
472: rc = ARGS_PHRASE;
473:
1.21 kristaps 474: /*
475: * Adjust new-buffer position to be beyond delimiter
476: * mark (e.g., Ta -> end + 2).
477: */
1.20 kristaps 478: if (p && pp) {
479: *pos += pp < p ? 2 : 1;
1.42 kristaps 480: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 481: p = pp < p ? pp : p;
482: } else if (p && ! pp) {
1.42 kristaps 483: rc = ARGS_PPHRASE;
1.20 kristaps 484: *pos += 1;
485: } else if (pp && ! p) {
1.17 kristaps 486: p = pp;
1.20 kristaps 487: *pos += 2;
1.44 kristaps 488: } else {
489: rc = ARGS_PEND;
1.20 kristaps 490: p = strchr(*v, 0);
1.44 kristaps 491: }
1.20 kristaps 492:
1.21 kristaps 493: /* Whitespace check for eoln case... */
1.52 kristaps 494: if ('\0' == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl))
1.49 kristaps 495: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
1.20 kristaps 496: return(ARGS_ERROR);
497:
498: *pos += (int)(p - *v);
1.17 kristaps 499:
500: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 501: pp = p - 1;
502: while (pp > *v && ' ' == *pp) {
503: if (pp > *v && '\\' == *(pp - 1))
504: break;
505: pp--;
1.17 kristaps 506: }
1.20 kristaps 507: *(pp + 1) = 0;
1.17 kristaps 508:
1.20 kristaps 509: /* Strip delimiter's proceeding whitespace. */
510: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
511: /* Skip ahead. */ ;
1.1 kristaps 512:
1.41 kristaps 513: return(rc);
1.17 kristaps 514: }
1.1 kristaps 515:
1.17 kristaps 516: /*
517: * Process a quoted literal. A quote begins with a double-quote
518: * and ends with a double-quote NOT preceded by a double-quote.
519: * Whitespace is NOT involved in literal termination.
520: */
1.1 kristaps 521:
1.46 kristaps 522: if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
523: if ( ! (MDOC_PHRASELIT & m->flags))
524: *v = &buf[++(*pos)];
525:
1.47 kristaps 526: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 527: m->flags |= MDOC_PHRASELIT;
1.1 kristaps 528:
1.17 kristaps 529: for ( ; buf[*pos]; (*pos)++) {
530: if ('\"' != buf[*pos])
531: continue;
532: if ('\"' != buf[*pos + 1])
533: break;
534: (*pos)++;
535: }
1.1 kristaps 536:
1.46 kristaps 537: if ('\0' == buf[*pos]) {
1.47 kristaps 538: if (ARGS_NOWARN & fl || MDOC_PPHRASE & m->flags)
1.23 kristaps 539: return(ARGS_QWORD);
1.49 kristaps 540: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE))
1.18 kristaps 541: return(ARGS_ERROR);
542: return(ARGS_QWORD);
1.17 kristaps 543: }
1.1 kristaps 544:
1.46 kristaps 545: m->flags &= ~MDOC_PHRASELIT;
546: buf[(*pos)++] = '\0';
1.1 kristaps 547:
1.46 kristaps 548: if ('\0' == buf[*pos])
1.17 kristaps 549: return(ARGS_QWORD);
1.1 kristaps 550:
1.17 kristaps 551: while (' ' == buf[*pos])
552: (*pos)++;
1.1 kristaps 553:
1.23 kristaps 554: if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
1.49 kristaps 555: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
1.17 kristaps 556: return(ARGS_ERROR);
1.1 kristaps 557:
1.17 kristaps 558: return(ARGS_QWORD);
1.1 kristaps 559: }
560:
1.17 kristaps 561: /*
562: * A non-quoted term progresses until either the end of line or
563: * a non-escaped whitespace.
1.1 kristaps 564: */
565:
1.17 kristaps 566: for ( ; buf[*pos]; (*pos)++)
1.52 kristaps 567: if (*pos && ' ' == buf[*pos] && '\\' != buf[*pos - 1])
1.17 kristaps 568: break;
1.1 kristaps 569:
1.46 kristaps 570: if ('\0' == buf[*pos])
1.17 kristaps 571: return(ARGS_WORD);
1.1 kristaps 572:
1.46 kristaps 573: buf[(*pos)++] = '\0';
1.1 kristaps 574:
1.17 kristaps 575: while (' ' == buf[*pos])
1.1 kristaps 576: (*pos)++;
577:
1.46 kristaps 578: if ('\0' == buf[*pos] && ! (ARGS_NOWARN & fl))
1.49 kristaps 579: if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE))
1.17 kristaps 580: return(ARGS_ERROR);
1.1 kristaps 581:
1.17 kristaps 582: return(ARGS_WORD);
1.1 kristaps 583: }
584:
585:
586: static int
1.35 kristaps 587: argv_a2arg(enum mdoct tok, const char *p)
1.1 kristaps 588: {
589:
590: /*
591: * Parse an argument identifier from its text. XXX - this
592: * should really be table-driven to clarify the code.
593: *
594: * If you add an argument to the list, make sure that you
595: * register it here with its one or more macros!
596: */
597:
598: switch (tok) {
599: case (MDOC_An):
1.27 kristaps 600: if (0 == strcmp(p, "split"))
1.1 kristaps 601: return(MDOC_Split);
1.27 kristaps 602: else if (0 == strcmp(p, "nosplit"))
1.1 kristaps 603: return(MDOC_Nosplit);
604: break;
605:
606: case (MDOC_Bd):
1.27 kristaps 607: if (0 == strcmp(p, "ragged"))
1.1 kristaps 608: return(MDOC_Ragged);
1.27 kristaps 609: else if (0 == strcmp(p, "unfilled"))
1.1 kristaps 610: return(MDOC_Unfilled);
1.27 kristaps 611: else if (0 == strcmp(p, "filled"))
1.1 kristaps 612: return(MDOC_Filled);
1.27 kristaps 613: else if (0 == strcmp(p, "literal"))
1.1 kristaps 614: return(MDOC_Literal);
1.27 kristaps 615: else if (0 == strcmp(p, "file"))
1.1 kristaps 616: return(MDOC_File);
1.27 kristaps 617: else if (0 == strcmp(p, "offset"))
1.1 kristaps 618: return(MDOC_Offset);
1.27 kristaps 619: else if (0 == strcmp(p, "compact"))
1.1 kristaps 620: return(MDOC_Compact);
1.28 kristaps 621: else if (0 == strcmp(p, "centered"))
622: return(MDOC_Centred);
1.1 kristaps 623: break;
624:
625: case (MDOC_Bf):
1.27 kristaps 626: if (0 == strcmp(p, "emphasis"))
1.1 kristaps 627: return(MDOC_Emphasis);
1.27 kristaps 628: else if (0 == strcmp(p, "literal"))
1.1 kristaps 629: return(MDOC_Literal);
1.27 kristaps 630: else if (0 == strcmp(p, "symbolic"))
1.1 kristaps 631: return(MDOC_Symbolic);
632: break;
633:
634: case (MDOC_Bk):
1.27 kristaps 635: if (0 == strcmp(p, "words"))
1.1 kristaps 636: return(MDOC_Words);
637: break;
638:
639: case (MDOC_Bl):
1.27 kristaps 640: if (0 == strcmp(p, "bullet"))
1.1 kristaps 641: return(MDOC_Bullet);
1.27 kristaps 642: else if (0 == strcmp(p, "dash"))
1.1 kristaps 643: return(MDOC_Dash);
1.27 kristaps 644: else if (0 == strcmp(p, "hyphen"))
1.1 kristaps 645: return(MDOC_Hyphen);
1.27 kristaps 646: else if (0 == strcmp(p, "item"))
1.1 kristaps 647: return(MDOC_Item);
1.27 kristaps 648: else if (0 == strcmp(p, "enum"))
1.1 kristaps 649: return(MDOC_Enum);
1.27 kristaps 650: else if (0 == strcmp(p, "tag"))
1.1 kristaps 651: return(MDOC_Tag);
1.27 kristaps 652: else if (0 == strcmp(p, "diag"))
1.1 kristaps 653: return(MDOC_Diag);
1.27 kristaps 654: else if (0 == strcmp(p, "hang"))
1.1 kristaps 655: return(MDOC_Hang);
1.27 kristaps 656: else if (0 == strcmp(p, "ohang"))
1.1 kristaps 657: return(MDOC_Ohang);
1.27 kristaps 658: else if (0 == strcmp(p, "inset"))
1.1 kristaps 659: return(MDOC_Inset);
1.27 kristaps 660: else if (0 == strcmp(p, "column"))
1.1 kristaps 661: return(MDOC_Column);
1.27 kristaps 662: else if (0 == strcmp(p, "width"))
1.1 kristaps 663: return(MDOC_Width);
1.27 kristaps 664: else if (0 == strcmp(p, "offset"))
1.1 kristaps 665: return(MDOC_Offset);
1.27 kristaps 666: else if (0 == strcmp(p, "compact"))
1.1 kristaps 667: return(MDOC_Compact);
1.27 kristaps 668: else if (0 == strcmp(p, "nested"))
1.1 kristaps 669: return(MDOC_Nested);
670: break;
671:
672: case (MDOC_Rv):
673: /* FALLTHROUGH */
674: case (MDOC_Ex):
1.27 kristaps 675: if (0 == strcmp(p, "std"))
1.1 kristaps 676: return(MDOC_Std);
677: break;
678: default:
679: break;
680: }
681:
682: return(MDOC_ARG_MAX);
683: }
684:
685:
686: static int
1.11 kristaps 687: argv_multi(struct mdoc *m, int line,
1.1 kristaps 688: struct mdoc_argv *v, int *pos, char *buf)
689: {
1.43 kristaps 690: enum margserr ac;
1.1 kristaps 691: char *p;
692:
693: for (v->sz = 0; ; v->sz++) {
694: if ('-' == buf[*pos])
695: break;
1.43 kristaps 696: ac = args(m, line, pos, buf, 0, &p);
697: if (ARGS_ERROR == ac)
1.1 kristaps 698: return(0);
1.43 kristaps 699: else if (ARGS_EOLN == ac)
1.1 kristaps 700: break;
701:
1.31 kristaps 702: if (0 == v->sz % MULTI_STEP)
703: v->value = mandoc_realloc(v->value,
1.1 kristaps 704: (v->sz + MULTI_STEP) * sizeof(char *));
1.31 kristaps 705:
706: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 707: }
708:
1.7 kristaps 709: return(1);
1.1 kristaps 710: }
711:
712:
713: static int
1.11 kristaps 714: argv_opt_single(struct mdoc *m, int line,
1.1 kristaps 715: struct mdoc_argv *v, int *pos, char *buf)
716: {
1.43 kristaps 717: enum margserr ac;
1.1 kristaps 718: char *p;
719:
720: if ('-' == buf[*pos])
721: return(1);
722:
1.43 kristaps 723: ac = args(m, line, pos, buf, 0, &p);
724: if (ARGS_ERROR == ac)
1.1 kristaps 725: return(0);
1.43 kristaps 726: if (ARGS_EOLN == ac)
1.1 kristaps 727: return(1);
728:
729: v->sz = 1;
1.31 kristaps 730: v->value = mandoc_malloc(sizeof(char *));
731: v->value[0] = mandoc_strdup(p);
1.2 kristaps 732:
1.1 kristaps 733: return(1);
734: }
735:
736:
737: /*
738: * Parse a single, mandatory value from the stream.
739: */
740: static int
1.11 kristaps 741: argv_single(struct mdoc *m, int line,
1.1 kristaps 742: struct mdoc_argv *v, int *pos, char *buf)
743: {
1.43 kristaps 744: int ppos;
745: enum margserr ac;
1.1 kristaps 746: char *p;
747:
748: ppos = *pos;
749:
1.43 kristaps 750: ac = args(m, line, pos, buf, 0, &p);
1.49 kristaps 751: if (ARGS_EOLN == ac) {
752: mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTARGVCOUNT);
753: return(0);
754: } else if (ARGS_ERROR == ac)
1.1 kristaps 755: return(0);
756:
757: v->sz = 1;
1.31 kristaps 758: v->value = mandoc_malloc(sizeof(char *));
759: v->value[0] = mandoc_strdup(p);
1.2 kristaps 760:
1.1 kristaps 761: return(1);
762: }
763:
764:
765: /*
766: * Determine rules for parsing arguments. Arguments can either accept
767: * no parameters, an optional single parameter, one parameter, or
768: * multiple parameters.
769: */
770: static int
771: argv(struct mdoc *mdoc, int line,
772: struct mdoc_argv *v, int *pos, char *buf)
773: {
774:
775: v->sz = 0;
776: v->value = NULL;
777:
778: switch (mdoc_argvflags[v->arg]) {
779: case (ARGV_SINGLE):
780: return(argv_single(mdoc, line, v, pos, buf));
781: case (ARGV_MULTI):
782: return(argv_multi(mdoc, line, v, pos, buf));
783: case (ARGV_OPT_SINGLE):
784: return(argv_opt_single(mdoc, line, v, pos, buf));
785: default:
786: /* ARGV_NONE */
787: break;
788: }
789:
790: return(1);
791: }
CVSweb