Annotation of mandoc/mdoc_argv.c, Revision 1.67
1.67 ! kristaps 1: /* $Id: mdoc_argv.c,v 1.66 2011/03/17 09:16:38 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)
1.67 ! kristaps 408: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.46 kristaps 409:
410: m->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 411: return(ARGS_EOLN);
1.46 kristaps 412: }
1.1 kristaps 413:
1.64 kristaps 414: *v = &buf[*pos];
1.17 kristaps 415:
1.64 kristaps 416: if (ARGS_DELIM & fl && args_checkpunct(&buf[*pos])) {
417: i = strlen(&buf[*pos]) + *pos;
418: if (i && ' ' != buf[i - 1])
419: return(ARGS_PUNCT);
420: if (ARGS_NOWARN & fl)
1.1 kristaps 421: return(ARGS_PUNCT);
1.67 ! kristaps 422: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.64 kristaps 423: return(ARGS_PUNCT);
1.1 kristaps 424: }
425:
1.17 kristaps 426: /*
427: * First handle TABSEP items, restricted to `Bl -column'. This
428: * ignores conventional token parsing and instead uses tabs or
429: * `Ta' macros to separate phrases. Phrases are parsed again
430: * for arguments at a later phase.
431: */
1.1 kristaps 432:
1.17 kristaps 433: if (ARGS_TABSEP & fl) {
1.20 kristaps 434: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 435: p = strchr(*v, '\t');
1.46 kristaps 436: pp = NULL;
1.17 kristaps 437:
438: /* Scan ahead to unescaped `Ta'. */
1.46 kristaps 439: if ( ! (MDOC_PHRASELIT & m->flags))
440: for (pp = *v; ; pp++) {
441: if (NULL == (pp = strstr(pp, "Ta")))
442: break;
443: if (pp > *v && ' ' != *(pp - 1))
444: continue;
1.51 kristaps 445: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 446: break;
447: }
1.17 kristaps 448:
1.42 kristaps 449: /* By default, assume a phrase. */
450: rc = ARGS_PHRASE;
451:
1.21 kristaps 452: /*
453: * Adjust new-buffer position to be beyond delimiter
454: * mark (e.g., Ta -> end + 2).
455: */
1.20 kristaps 456: if (p && pp) {
457: *pos += pp < p ? 2 : 1;
1.42 kristaps 458: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 459: p = pp < p ? pp : p;
460: } else if (p && ! pp) {
1.42 kristaps 461: rc = ARGS_PPHRASE;
1.20 kristaps 462: *pos += 1;
463: } else if (pp && ! p) {
1.17 kristaps 464: p = pp;
1.20 kristaps 465: *pos += 2;
1.44 kristaps 466: } else {
467: rc = ARGS_PEND;
1.20 kristaps 468: p = strchr(*v, 0);
1.44 kristaps 469: }
1.20 kristaps 470:
1.21 kristaps 471: /* Whitespace check for eoln case... */
1.52 kristaps 472: if ('\0' == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl))
1.67 ! kristaps 473: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.20 kristaps 474:
475: *pos += (int)(p - *v);
1.17 kristaps 476:
477: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 478: pp = p - 1;
479: while (pp > *v && ' ' == *pp) {
480: if (pp > *v && '\\' == *(pp - 1))
481: break;
482: pp--;
1.17 kristaps 483: }
1.20 kristaps 484: *(pp + 1) = 0;
1.17 kristaps 485:
1.20 kristaps 486: /* Strip delimiter's proceeding whitespace. */
487: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
488: /* Skip ahead. */ ;
1.1 kristaps 489:
1.41 kristaps 490: return(rc);
1.17 kristaps 491: }
1.1 kristaps 492:
1.17 kristaps 493: /*
494: * Process a quoted literal. A quote begins with a double-quote
495: * and ends with a double-quote NOT preceded by a double-quote.
496: * Whitespace is NOT involved in literal termination.
497: */
1.1 kristaps 498:
1.46 kristaps 499: if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
500: if ( ! (MDOC_PHRASELIT & m->flags))
501: *v = &buf[++(*pos)];
502:
1.47 kristaps 503: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 504: m->flags |= MDOC_PHRASELIT;
1.1 kristaps 505:
1.17 kristaps 506: for ( ; buf[*pos]; (*pos)++) {
507: if ('\"' != buf[*pos])
508: continue;
509: if ('\"' != buf[*pos + 1])
510: break;
511: (*pos)++;
512: }
1.1 kristaps 513:
1.46 kristaps 514: if ('\0' == buf[*pos]) {
1.47 kristaps 515: if (ARGS_NOWARN & fl || MDOC_PPHRASE & m->flags)
1.23 kristaps 516: return(ARGS_QWORD);
1.67 ! kristaps 517: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.18 kristaps 518: return(ARGS_QWORD);
1.17 kristaps 519: }
1.1 kristaps 520:
1.46 kristaps 521: m->flags &= ~MDOC_PHRASELIT;
522: buf[(*pos)++] = '\0';
1.1 kristaps 523:
1.46 kristaps 524: if ('\0' == buf[*pos])
1.17 kristaps 525: return(ARGS_QWORD);
1.1 kristaps 526:
1.17 kristaps 527: while (' ' == buf[*pos])
528: (*pos)++;
1.1 kristaps 529:
1.23 kristaps 530: if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
1.67 ! kristaps 531: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 532:
1.17 kristaps 533: return(ARGS_QWORD);
1.1 kristaps 534: }
535:
1.17 kristaps 536: /*
537: * A non-quoted term progresses until either the end of line or
538: * a non-escaped whitespace.
1.1 kristaps 539: */
540:
1.17 kristaps 541: for ( ; buf[*pos]; (*pos)++)
1.52 kristaps 542: if (*pos && ' ' == buf[*pos] && '\\' != buf[*pos - 1])
1.17 kristaps 543: break;
1.1 kristaps 544:
1.46 kristaps 545: if ('\0' == buf[*pos])
1.17 kristaps 546: return(ARGS_WORD);
1.1 kristaps 547:
1.46 kristaps 548: buf[(*pos)++] = '\0';
1.1 kristaps 549:
1.17 kristaps 550: while (' ' == buf[*pos])
1.1 kristaps 551: (*pos)++;
552:
1.46 kristaps 553: if ('\0' == buf[*pos] && ! (ARGS_NOWARN & fl))
1.67 ! kristaps 554: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 555:
1.17 kristaps 556: return(ARGS_WORD);
1.64 kristaps 557: }
558:
559: /*
560: * Check if the string consists only of space-separated closing
1.65 kristaps 561: * delimiters. This is a bit of a dance: the first must be a close
562: * delimiter, but it may be followed by middle delimiters. Arbitrary
563: * whitespace may separate these tokens.
1.64 kristaps 564: */
565: static int
566: args_checkpunct(const char *p)
567: {
1.65 kristaps 568: int i, j;
569: char buf[DELIMSZ];
1.64 kristaps 570: enum mdelim d;
571:
572: i = 0;
573:
1.65 kristaps 574: /* First token must be a close-delimiter. */
575:
576: for (j = 0; p[i] && ' ' != p[i] && j < DELIMSZ; j++, i++)
577: buf[j] = p[i];
578:
579: if (DELIMSZ == j)
580: return(0);
581:
582: buf[j] = '\0';
1.66 kristaps 583: if (DELIM_CLOSE != mandoc_isdelim(buf))
1.64 kristaps 584: return(0);
585:
1.65 kristaps 586: while (' ' == p[i])
587: i++;
588:
589: /* Remaining must NOT be open/none. */
590:
591: while (p[i]) {
592: j = 0;
593: while (p[i] && ' ' != p[i] && j < DELIMSZ)
594: buf[j++] = p[i++];
595:
596: if (DELIMSZ == j)
597: return(0);
598:
599: buf[j] = '\0';
1.66 kristaps 600: d = mandoc_isdelim(buf);
1.64 kristaps 601: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 602: return(0);
603:
604: while (' ' == p[i])
1.64 kristaps 605: i++;
606: }
607:
608: return('\0' == p[i]);
1.1 kristaps 609: }
610:
1.63 kristaps 611: /*
612: * Match up an argument string (e.g., `-foo bar' having "foo") with the
613: * correrct identifier. It must apply to the given macro. If none was
614: * found (including bad matches), return MDOC_ARG_MAX.
615: */
1.57 kristaps 616: static enum mdocargt
1.35 kristaps 617: argv_a2arg(enum mdoct tok, const char *p)
1.1 kristaps 618: {
1.63 kristaps 619: enum mdocargt args[MDOC_ARG_MAX];
620: int i, len;
1.1 kristaps 621:
1.63 kristaps 622: len = 0;
1.1 kristaps 623:
624: switch (tok) {
625: case (MDOC_An):
1.63 kristaps 626: args[len++] = MDOC_Split;
627: args[len++] = MDOC_Nosplit;
1.1 kristaps 628: break;
629: case (MDOC_Bd):
1.63 kristaps 630: args[len++] = MDOC_Ragged;
631: args[len++] = MDOC_Unfilled;
632: args[len++] = MDOC_Filled;
633: args[len++] = MDOC_Literal;
634: args[len++] = MDOC_File;
635: args[len++] = MDOC_Offset;
636: args[len++] = MDOC_Compact;
637: args[len++] = MDOC_Centred;
1.1 kristaps 638: break;
639: case (MDOC_Bf):
1.63 kristaps 640: args[len++] = MDOC_Emphasis;
641: args[len++] = MDOC_Literal;
642: args[len++] = MDOC_Symbolic;
1.1 kristaps 643: break;
644: case (MDOC_Bk):
1.63 kristaps 645: args[len++] = MDOC_Words;
1.1 kristaps 646: break;
647: case (MDOC_Bl):
1.63 kristaps 648: args[len++] = MDOC_Bullet;
649: args[len++] = MDOC_Dash;
650: args[len++] = MDOC_Hyphen;
651: args[len++] = MDOC_Item;
652: args[len++] = MDOC_Enum;
653: args[len++] = MDOC_Tag;
654: args[len++] = MDOC_Diag;
655: args[len++] = MDOC_Hang;
656: args[len++] = MDOC_Ohang;
657: args[len++] = MDOC_Inset;
658: args[len++] = MDOC_Column;
659: args[len++] = MDOC_Width;
660: args[len++] = MDOC_Offset;
661: args[len++] = MDOC_Compact;
662: args[len++] = MDOC_Nested;
1.1 kristaps 663: break;
664: case (MDOC_Rv):
665: /* FALLTHROUGH */
666: case (MDOC_Ex):
1.63 kristaps 667: args[len++] = MDOC_Std;
1.1 kristaps 668: break;
669: default:
670: break;
671: }
1.63 kristaps 672:
673: for (i = 0; i < len; i++)
674: if (0 == strcmp(p, mdoc_argnames[args[i]]))
675: return(args[i]);
1.1 kristaps 676:
677: return(MDOC_ARG_MAX);
678: }
679:
680:
681: static int
1.11 kristaps 682: argv_multi(struct mdoc *m, int line,
1.1 kristaps 683: struct mdoc_argv *v, int *pos, char *buf)
684: {
1.43 kristaps 685: enum margserr ac;
1.1 kristaps 686: char *p;
687:
688: for (v->sz = 0; ; v->sz++) {
689: if ('-' == buf[*pos])
690: break;
1.43 kristaps 691: ac = args(m, line, pos, buf, 0, &p);
692: if (ARGS_ERROR == ac)
1.1 kristaps 693: return(0);
1.43 kristaps 694: else if (ARGS_EOLN == ac)
1.1 kristaps 695: break;
696:
1.31 kristaps 697: if (0 == v->sz % MULTI_STEP)
698: v->value = mandoc_realloc(v->value,
1.1 kristaps 699: (v->sz + MULTI_STEP) * sizeof(char *));
1.31 kristaps 700:
701: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 702: }
703:
1.7 kristaps 704: return(1);
1.1 kristaps 705: }
706:
707:
708: static int
1.11 kristaps 709: argv_opt_single(struct mdoc *m, int line,
1.1 kristaps 710: struct mdoc_argv *v, int *pos, char *buf)
711: {
1.43 kristaps 712: enum margserr ac;
1.1 kristaps 713: char *p;
714:
715: if ('-' == buf[*pos])
716: return(1);
717:
1.43 kristaps 718: ac = args(m, line, pos, buf, 0, &p);
719: if (ARGS_ERROR == ac)
1.1 kristaps 720: return(0);
1.43 kristaps 721: if (ARGS_EOLN == ac)
1.1 kristaps 722: return(1);
723:
724: v->sz = 1;
1.31 kristaps 725: v->value = mandoc_malloc(sizeof(char *));
726: v->value[0] = mandoc_strdup(p);
1.2 kristaps 727:
1.1 kristaps 728: return(1);
729: }
730:
731:
732: /*
733: * Parse a single, mandatory value from the stream.
734: */
735: static int
1.11 kristaps 736: argv_single(struct mdoc *m, int line,
1.1 kristaps 737: struct mdoc_argv *v, int *pos, char *buf)
738: {
1.43 kristaps 739: int ppos;
740: enum margserr ac;
1.1 kristaps 741: char *p;
742:
743: ppos = *pos;
744:
1.43 kristaps 745: ac = args(m, line, pos, buf, 0, &p);
1.49 kristaps 746: if (ARGS_EOLN == ac) {
747: mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTARGVCOUNT);
748: return(0);
749: } else if (ARGS_ERROR == ac)
1.1 kristaps 750: return(0);
751:
752: v->sz = 1;
1.31 kristaps 753: v->value = mandoc_malloc(sizeof(char *));
754: v->value[0] = mandoc_strdup(p);
1.2 kristaps 755:
1.1 kristaps 756: return(1);
757: }
758:
759:
760: /*
761: * Determine rules for parsing arguments. Arguments can either accept
762: * no parameters, an optional single parameter, one parameter, or
763: * multiple parameters.
764: */
765: static int
766: argv(struct mdoc *mdoc, int line,
767: struct mdoc_argv *v, int *pos, char *buf)
768: {
769:
770: v->sz = 0;
771: v->value = NULL;
772:
773: switch (mdoc_argvflags[v->arg]) {
774: case (ARGV_SINGLE):
775: return(argv_single(mdoc, line, v, pos, buf));
776: case (ARGV_MULTI):
777: return(argv_multi(mdoc, line, v, pos, buf));
778: case (ARGV_OPT_SINGLE):
779: return(argv_opt_single(mdoc, line, v, pos, buf));
780: default:
781: /* ARGV_NONE */
782: break;
783: }
784:
785: return(1);
786: }
CVSweb