Annotation of mandoc/mdoc_argv.c, Revision 1.70
1.70 ! kristaps 1: /* $Id: mdoc_argv.c,v 1.69 2011/03/17 11:09:36 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:
1.68 kristaps 33: #define MULTI_STEP 5 /* pre-allocate argument values */
1.1 kristaps 34:
1.57 kristaps 35: static enum mdocargt argv_a2arg(enum mdoct, const char *);
1.39 kristaps 36: static enum margserr args(struct mdoc *, int, int *,
1.1 kristaps 37: char *, int, char **);
1.69 kristaps 38: static int args_checkpunct(struct mdoc *,
39: const char *, int, int, int);
1.1 kristaps 40: static int argv(struct mdoc *, int,
41: struct mdoc_argv *, int *, char *);
42: static int argv_single(struct mdoc *, int,
43: struct mdoc_argv *, int *, char *);
44: static int argv_opt_single(struct mdoc *, int,
45: struct mdoc_argv *, int *, char *);
46: static int argv_multi(struct mdoc *, int,
47: struct mdoc_argv *, int *, char *);
1.2 kristaps 48:
1.68 kristaps 49: enum argvflag {
50: ARGV_NONE, /* no args to flag (e.g., -split) */
51: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
52: ARGV_MULTI, /* multiple args (e.g., -column xxx yyy) */
53: ARGV_OPT_SINGLE /* optional arg (e.g., -offset [xxx]) */
54: };
1.1 kristaps 55:
1.68 kristaps 56: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 57: ARGV_NONE, /* MDOC_Split */
58: ARGV_NONE, /* MDOC_Nosplit */
59: ARGV_NONE, /* MDOC_Ragged */
60: ARGV_NONE, /* MDOC_Unfilled */
61: ARGV_NONE, /* MDOC_Literal */
1.29 kristaps 62: ARGV_SINGLE, /* MDOC_File */
1.26 kristaps 63: ARGV_OPT_SINGLE, /* MDOC_Offset */
1.1 kristaps 64: ARGV_NONE, /* MDOC_Bullet */
65: ARGV_NONE, /* MDOC_Dash */
66: ARGV_NONE, /* MDOC_Hyphen */
67: ARGV_NONE, /* MDOC_Item */
68: ARGV_NONE, /* MDOC_Enum */
69: ARGV_NONE, /* MDOC_Tag */
70: ARGV_NONE, /* MDOC_Diag */
71: ARGV_NONE, /* MDOC_Hang */
72: ARGV_NONE, /* MDOC_Ohang */
73: ARGV_NONE, /* MDOC_Inset */
74: ARGV_MULTI, /* MDOC_Column */
75: ARGV_SINGLE, /* MDOC_Width */
76: ARGV_NONE, /* MDOC_Compact */
1.25 kristaps 77: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 78: ARGV_NONE, /* MDOC_Filled */
79: ARGV_NONE, /* MDOC_Words */
80: ARGV_NONE, /* MDOC_Emphasis */
81: ARGV_NONE, /* MDOC_Symbolic */
82: ARGV_NONE /* MDOC_Symbolic */
83: };
84:
1.68 kristaps 85: static const int argflags[MDOC_MAX] = {
1.5 kristaps 86: 0, /* Ap */
1.1 kristaps 87: 0, /* Dd */
88: 0, /* Dt */
89: 0, /* Os */
1.17 kristaps 90: 0, /* Sh */
91: 0, /* Ss */
1.59 kristaps 92: 0, /* Pp */
1.1 kristaps 93: ARGS_DELIM, /* D1 */
1.17 kristaps 94: ARGS_DELIM, /* Dl */
1.1 kristaps 95: 0, /* Bd */
96: 0, /* Ed */
1.17 kristaps 97: 0, /* Bl */
1.1 kristaps 98: 0, /* El */
99: 0, /* It */
100: ARGS_DELIM, /* Ad */
101: ARGS_DELIM, /* An */
1.17 kristaps 102: ARGS_DELIM, /* Ar */
103: 0, /* Cd */
1.1 kristaps 104: ARGS_DELIM, /* Cm */
105: ARGS_DELIM, /* Dv */
106: ARGS_DELIM, /* Er */
107: ARGS_DELIM, /* Ev */
108: 0, /* Ex */
1.17 kristaps 109: ARGS_DELIM, /* Fa */
1.1 kristaps 110: 0, /* Fd */
111: ARGS_DELIM, /* Fl */
1.17 kristaps 112: ARGS_DELIM, /* Fn */
113: ARGS_DELIM, /* Ft */
1.1 kristaps 114: ARGS_DELIM, /* Ic */
115: 0, /* In */
1.17 kristaps 116: ARGS_DELIM, /* Li */
117: 0, /* Nd */
1.1 kristaps 118: ARGS_DELIM, /* Nm */
119: ARGS_DELIM, /* Op */
120: 0, /* Ot */
121: ARGS_DELIM, /* Pa */
122: 0, /* Rv */
1.15 kristaps 123: ARGS_DELIM, /* St */
1.1 kristaps 124: ARGS_DELIM, /* Va */
125: ARGS_DELIM, /* Vt */
126: ARGS_DELIM, /* Xr */
1.17 kristaps 127: 0, /* %A */
128: 0, /* %B */
129: 0, /* %D */
130: 0, /* %I */
131: 0, /* %J */
132: 0, /* %N */
133: 0, /* %O */
134: 0, /* %P */
135: 0, /* %R */
136: 0, /* %T */
137: 0, /* %V */
1.1 kristaps 138: ARGS_DELIM, /* Ac */
139: 0, /* Ao */
140: ARGS_DELIM, /* Aq */
141: ARGS_DELIM, /* At */
142: ARGS_DELIM, /* Bc */
143: 0, /* Bf */
144: 0, /* Bo */
145: ARGS_DELIM, /* Bq */
146: ARGS_DELIM, /* Bsx */
147: ARGS_DELIM, /* Bx */
148: 0, /* Db */
149: ARGS_DELIM, /* Dc */
150: 0, /* Do */
151: ARGS_DELIM, /* Dq */
152: ARGS_DELIM, /* Ec */
153: 0, /* Ef */
154: ARGS_DELIM, /* Em */
155: 0, /* Eo */
156: ARGS_DELIM, /* Fx */
157: ARGS_DELIM, /* Ms */
158: ARGS_DELIM, /* No */
159: ARGS_DELIM, /* Ns */
160: ARGS_DELIM, /* Nx */
161: ARGS_DELIM, /* Ox */
162: ARGS_DELIM, /* Pc */
163: ARGS_DELIM, /* Pf */
164: 0, /* Po */
165: ARGS_DELIM, /* Pq */
166: ARGS_DELIM, /* Qc */
167: ARGS_DELIM, /* Ql */
168: 0, /* Qo */
169: ARGS_DELIM, /* Qq */
170: 0, /* Re */
171: 0, /* Rs */
172: ARGS_DELIM, /* Sc */
173: 0, /* So */
174: ARGS_DELIM, /* Sq */
175: 0, /* Sm */
176: ARGS_DELIM, /* Sx */
1.17 kristaps 177: ARGS_DELIM, /* Sy */
1.1 kristaps 178: ARGS_DELIM, /* Tn */
179: ARGS_DELIM, /* Ux */
180: ARGS_DELIM, /* Xc */
181: 0, /* Xo */
1.17 kristaps 182: 0, /* Fo */
1.1 kristaps 183: 0, /* Fc */
184: 0, /* Oo */
185: ARGS_DELIM, /* Oc */
186: 0, /* Bk */
187: 0, /* Ek */
188: 0, /* Bt */
189: 0, /* Hf */
190: 0, /* Fr */
191: 0, /* Ud */
192: 0, /* Lb */
1.60 schwarze 193: 0, /* Lp */
1.17 kristaps 194: ARGS_DELIM, /* Lk */
195: ARGS_DELIM, /* Mt */
1.1 kristaps 196: ARGS_DELIM, /* Brq */
197: 0, /* Bro */
198: ARGS_DELIM, /* Brc */
1.17 kristaps 199: 0, /* %C */
1.1 kristaps 200: 0, /* Es */
201: 0, /* En */
202: 0, /* Dx */
1.17 kristaps 203: 0, /* %Q */
1.14 kristaps 204: 0, /* br */
205: 0, /* sp */
1.30 kristaps 206: 0, /* %U */
1.51 kristaps 207: 0, /* Ta */
1.1 kristaps 208: };
209:
210: /*
211: * Parse an argument from line text. This comes in the form of -key
212: * [value0...], which may either have a single mandatory value, at least
213: * one mandatory value, an optional single value, or no value.
214: */
1.40 kristaps 215: enum margverr
1.35 kristaps 216: mdoc_argv(struct mdoc *m, int line, enum mdoct tok,
1.1 kristaps 217: struct mdoc_arg **v, int *pos, char *buf)
218: {
219: char *p, sv;
220: struct mdoc_argv tmp;
221: struct mdoc_arg *arg;
222:
1.52 kristaps 223: if ('\0' == buf[*pos])
1.1 kristaps 224: return(ARGV_EOLN);
225:
226: assert(' ' != buf[*pos]);
227:
228: /* Parse through to the first unescaped space. */
229:
230: p = &buf[++(*pos)];
231:
232: assert(*pos > 0);
233:
234: /* LINTED */
235: while (buf[*pos]) {
236: if (' ' == buf[*pos])
237: if ('\\' != buf[*pos - 1])
238: break;
239: (*pos)++;
240: }
241:
242: /* XXX - save zeroed byte, if not an argument. */
243:
1.52 kristaps 244: sv = '\0';
1.1 kristaps 245: if (buf[*pos]) {
246: sv = buf[*pos];
1.52 kristaps 247: buf[(*pos)++] = '\0';
1.1 kristaps 248: }
249:
1.68 kristaps 250: memset(&tmp, 0, sizeof(struct mdoc_argv));
1.1 kristaps 251: tmp.line = line;
252: tmp.pos = *pos;
253:
254: /* See if our token accepts the argument. */
255:
256: if (MDOC_ARG_MAX == (tmp.arg = argv_a2arg(tok, p))) {
257: /* XXX - restore saved zeroed byte. */
258: if (sv)
259: buf[*pos - 1] = sv;
260: return(ARGV_WORD);
261: }
262:
263: while (buf[*pos] && ' ' == buf[*pos])
264: (*pos)++;
265:
1.11 kristaps 266: if ( ! argv(m, line, &tmp, pos, buf))
1.1 kristaps 267: return(ARGV_ERROR);
268:
1.31 kristaps 269: if (NULL == (arg = *v))
1.32 kristaps 270: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 271:
272: arg->argc++;
1.31 kristaps 273: arg->argv = mandoc_realloc
274: (arg->argv, arg->argc * sizeof(struct mdoc_argv));
1.2 kristaps 275:
1.68 kristaps 276: memcpy(&arg->argv[(int)arg->argc - 1],
1.1 kristaps 277: &tmp, sizeof(struct mdoc_argv));
278:
279: return(ARGV_ARG);
280: }
281:
282: void
283: mdoc_argv_free(struct mdoc_arg *p)
284: {
1.37 kristaps 285: int i;
1.1 kristaps 286:
287: if (NULL == p)
288: return;
289:
290: if (p->refcnt) {
291: --(p->refcnt);
292: if (p->refcnt)
293: return;
294: }
295: assert(p->argc);
296:
1.37 kristaps 297: for (i = (int)p->argc - 1; i >= 0; i--)
298: mdoc_argn_free(p, i);
299:
300: free(p->argv);
301: free(p);
302: }
303:
304: void
305: mdoc_argn_free(struct mdoc_arg *p, int iarg)
306: {
1.58 kristaps 307: struct mdoc_argv *arg;
1.37 kristaps 308: int j;
1.58 kristaps 309:
310: arg = &p->argv[iarg];
1.1 kristaps 311:
1.37 kristaps 312: if (arg->sz && arg->value) {
313: for (j = (int)arg->sz - 1; j >= 0; j--)
314: free(arg->value[j]);
315: free(arg->value);
1.1 kristaps 316: }
317:
1.37 kristaps 318: for (--p->argc; iarg < (int)p->argc; iarg++)
319: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 320: }
321:
1.39 kristaps 322: enum margserr
1.23 kristaps 323: mdoc_zargs(struct mdoc *m, int line, int *pos,
324: char *buf, int flags, char **v)
1.19 kristaps 325: {
326:
1.23 kristaps 327: return(args(m, line, pos, buf, flags, v));
1.19 kristaps 328: }
329:
1.39 kristaps 330: enum margserr
1.35 kristaps 331: mdoc_args(struct mdoc *m, int line, int *pos,
332: char *buf, enum mdoct tok, char **v)
1.1 kristaps 333: {
1.50 kristaps 334: int fl;
1.1 kristaps 335: struct mdoc_node *n;
336:
1.68 kristaps 337: fl = argflags[tok];
1.1 kristaps 338:
1.17 kristaps 339: if (MDOC_It != tok)
340: return(args(m, line, pos, buf, fl, v));
341:
1.51 kristaps 342: /*
343: * We know that we're in an `It', so it's reasonable to expect
344: * us to be sitting in a `Bl'. Someday this may not be the case
345: * (if we allow random `It's sitting out there), so provide a
346: * safe fall-back into the default behaviour.
347: */
348:
1.17 kristaps 349: for (n = m->last; n; n = n->parent)
1.50 kristaps 350: if (MDOC_Bl == n->tok)
1.17 kristaps 351: break;
1.1 kristaps 352:
1.62 kristaps 353: if (n && LIST_column == n->norm->Bl.type) {
1.17 kristaps 354: fl |= ARGS_TABSEP;
355: fl &= ~ARGS_DELIM;
1.1 kristaps 356: }
357:
1.11 kristaps 358: return(args(m, line, pos, buf, fl, v));
1.1 kristaps 359: }
360:
1.39 kristaps 361: static enum margserr
1.13 kristaps 362: args(struct mdoc *m, int line, int *pos,
363: char *buf, int fl, char **v)
1.1 kristaps 364: {
1.52 kristaps 365: char *p, *pp;
366: enum margserr rc;
1.1 kristaps 367:
1.24 kristaps 368: /*
369: * Parse out the terms (like `val' in `.Xx -arg val' or simply
370: * `.Xx val'), which can have all sorts of properties:
371: *
372: * ARGS_DELIM: use special handling if encountering trailing
373: * delimiters in the form of [[::delim::][ ]+]+.
374: *
375: * ARGS_NOWARN: don't post warnings. This is only used when
376: * re-parsing delimiters, as the warnings have already been
377: * posted.
378: *
379: * ARGS_TABSEP: use special handling for tab/`Ta' separated
380: * phrases like in `Bl -column'.
381: */
382:
1.17 kristaps 383: assert(' ' != buf[*pos]);
1.1 kristaps 384:
1.46 kristaps 385: if ('\0' == buf[*pos]) {
1.47 kristaps 386: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 387: return(ARGS_EOLN);
388: /*
389: * If we're not in a partial phrase and the flag for
390: * being a phrase literal is still set, the punctuation
391: * is unterminated.
392: */
393: if (MDOC_PHRASELIT & m->flags)
1.67 kristaps 394: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.46 kristaps 395:
396: m->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 397: return(ARGS_EOLN);
1.46 kristaps 398: }
1.1 kristaps 399:
1.64 kristaps 400: *v = &buf[*pos];
1.17 kristaps 401:
1.69 kristaps 402: if (ARGS_DELIM & fl && args_checkpunct(m, buf, *pos, line, fl))
1.64 kristaps 403: return(ARGS_PUNCT);
1.1 kristaps 404:
1.17 kristaps 405: /*
406: * First handle TABSEP items, restricted to `Bl -column'. This
407: * ignores conventional token parsing and instead uses tabs or
408: * `Ta' macros to separate phrases. Phrases are parsed again
409: * for arguments at a later phase.
410: */
1.1 kristaps 411:
1.17 kristaps 412: if (ARGS_TABSEP & fl) {
1.20 kristaps 413: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 414: p = strchr(*v, '\t');
1.46 kristaps 415: pp = NULL;
1.17 kristaps 416:
417: /* Scan ahead to unescaped `Ta'. */
1.46 kristaps 418: if ( ! (MDOC_PHRASELIT & m->flags))
419: for (pp = *v; ; pp++) {
420: if (NULL == (pp = strstr(pp, "Ta")))
421: break;
422: if (pp > *v && ' ' != *(pp - 1))
423: continue;
1.51 kristaps 424: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 425: break;
426: }
1.17 kristaps 427:
1.42 kristaps 428: /* By default, assume a phrase. */
429: rc = ARGS_PHRASE;
430:
1.21 kristaps 431: /*
432: * Adjust new-buffer position to be beyond delimiter
433: * mark (e.g., Ta -> end + 2).
434: */
1.20 kristaps 435: if (p && pp) {
436: *pos += pp < p ? 2 : 1;
1.42 kristaps 437: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 438: p = pp < p ? pp : p;
439: } else if (p && ! pp) {
1.42 kristaps 440: rc = ARGS_PPHRASE;
1.20 kristaps 441: *pos += 1;
442: } else if (pp && ! p) {
1.17 kristaps 443: p = pp;
1.20 kristaps 444: *pos += 2;
1.44 kristaps 445: } else {
446: rc = ARGS_PEND;
1.20 kristaps 447: p = strchr(*v, 0);
1.44 kristaps 448: }
1.20 kristaps 449:
1.21 kristaps 450: /* Whitespace check for eoln case... */
1.52 kristaps 451: if ('\0' == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl))
1.67 kristaps 452: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.20 kristaps 453:
454: *pos += (int)(p - *v);
1.17 kristaps 455:
456: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 457: pp = p - 1;
458: while (pp > *v && ' ' == *pp) {
459: if (pp > *v && '\\' == *(pp - 1))
460: break;
461: pp--;
1.17 kristaps 462: }
1.20 kristaps 463: *(pp + 1) = 0;
1.17 kristaps 464:
1.20 kristaps 465: /* Strip delimiter's proceeding whitespace. */
466: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
467: /* Skip ahead. */ ;
1.1 kristaps 468:
1.41 kristaps 469: return(rc);
1.17 kristaps 470: }
1.1 kristaps 471:
1.17 kristaps 472: /*
473: * Process a quoted literal. A quote begins with a double-quote
474: * and ends with a double-quote NOT preceded by a double-quote.
475: * Whitespace is NOT involved in literal termination.
476: */
1.1 kristaps 477:
1.46 kristaps 478: if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
479: if ( ! (MDOC_PHRASELIT & m->flags))
480: *v = &buf[++(*pos)];
481:
1.47 kristaps 482: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 483: m->flags |= MDOC_PHRASELIT;
1.1 kristaps 484:
1.17 kristaps 485: for ( ; buf[*pos]; (*pos)++) {
486: if ('\"' != buf[*pos])
487: continue;
488: if ('\"' != buf[*pos + 1])
489: break;
490: (*pos)++;
491: }
1.1 kristaps 492:
1.46 kristaps 493: if ('\0' == buf[*pos]) {
1.47 kristaps 494: if (ARGS_NOWARN & fl || MDOC_PPHRASE & m->flags)
1.23 kristaps 495: return(ARGS_QWORD);
1.67 kristaps 496: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.18 kristaps 497: return(ARGS_QWORD);
1.17 kristaps 498: }
1.1 kristaps 499:
1.46 kristaps 500: m->flags &= ~MDOC_PHRASELIT;
501: buf[(*pos)++] = '\0';
1.1 kristaps 502:
1.46 kristaps 503: if ('\0' == buf[*pos])
1.17 kristaps 504: return(ARGS_QWORD);
1.1 kristaps 505:
1.17 kristaps 506: while (' ' == buf[*pos])
507: (*pos)++;
1.1 kristaps 508:
1.23 kristaps 509: if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
1.67 kristaps 510: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 511:
1.17 kristaps 512: return(ARGS_QWORD);
1.1 kristaps 513: }
514:
1.17 kristaps 515: /*
516: * A non-quoted term progresses until either the end of line or
517: * a non-escaped whitespace.
1.1 kristaps 518: */
519:
1.17 kristaps 520: for ( ; buf[*pos]; (*pos)++)
1.52 kristaps 521: if (*pos && ' ' == buf[*pos] && '\\' != buf[*pos - 1])
1.17 kristaps 522: break;
1.1 kristaps 523:
1.46 kristaps 524: if ('\0' == buf[*pos])
1.17 kristaps 525: return(ARGS_WORD);
1.1 kristaps 526:
1.46 kristaps 527: buf[(*pos)++] = '\0';
1.1 kristaps 528:
1.17 kristaps 529: while (' ' == buf[*pos])
1.1 kristaps 530: (*pos)++;
531:
1.46 kristaps 532: if ('\0' == buf[*pos] && ! (ARGS_NOWARN & fl))
1.67 kristaps 533: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 534:
1.17 kristaps 535: return(ARGS_WORD);
1.64 kristaps 536: }
537:
538: /*
539: * Check if the string consists only of space-separated closing
1.65 kristaps 540: * delimiters. This is a bit of a dance: the first must be a close
541: * delimiter, but it may be followed by middle delimiters. Arbitrary
542: * whitespace may separate these tokens.
1.64 kristaps 543: */
544: static int
1.69 kristaps 545: args_checkpunct(struct mdoc *m, const char *buf, int i, int ln, int fl)
1.64 kristaps 546: {
1.69 kristaps 547: int j;
548: char dbuf[DELIMSZ];
1.64 kristaps 549: enum mdelim d;
550:
1.65 kristaps 551: /* First token must be a close-delimiter. */
552:
1.69 kristaps 553: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
554: dbuf[j] = buf[i];
1.65 kristaps 555:
556: if (DELIMSZ == j)
557: return(0);
558:
1.69 kristaps 559: dbuf[j] = '\0';
1.70 ! kristaps 560: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.64 kristaps 561: return(0);
562:
1.69 kristaps 563: while (' ' == buf[i])
1.65 kristaps 564: i++;
565:
566: /* Remaining must NOT be open/none. */
567:
1.69 kristaps 568: while (buf[i]) {
1.65 kristaps 569: j = 0;
1.69 kristaps 570: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
571: dbuf[j++] = buf[i++];
1.65 kristaps 572:
573: if (DELIMSZ == j)
574: return(0);
575:
1.69 kristaps 576: dbuf[j] = '\0';
1.70 ! kristaps 577: d = mdoc_isdelim(dbuf);
1.64 kristaps 578: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 579: return(0);
580:
1.69 kristaps 581: while (' ' == buf[i])
1.64 kristaps 582: i++;
583: }
584:
1.69 kristaps 585: if ( ! (ARGS_NOWARN & fl) && i && ' ' == buf[i - 1])
586: mdoc_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE);
587:
588: return('\0' == buf[i]);
1.1 kristaps 589: }
590:
1.63 kristaps 591: /*
592: * Match up an argument string (e.g., `-foo bar' having "foo") with the
593: * correrct identifier. It must apply to the given macro. If none was
594: * found (including bad matches), return MDOC_ARG_MAX.
595: */
1.57 kristaps 596: static enum mdocargt
1.35 kristaps 597: argv_a2arg(enum mdoct tok, const char *p)
1.1 kristaps 598: {
1.63 kristaps 599: enum mdocargt args[MDOC_ARG_MAX];
600: int i, len;
1.1 kristaps 601:
1.63 kristaps 602: len = 0;
1.1 kristaps 603:
604: switch (tok) {
605: case (MDOC_An):
1.63 kristaps 606: args[len++] = MDOC_Split;
607: args[len++] = MDOC_Nosplit;
1.1 kristaps 608: break;
609: case (MDOC_Bd):
1.63 kristaps 610: args[len++] = MDOC_Ragged;
611: args[len++] = MDOC_Unfilled;
612: args[len++] = MDOC_Filled;
613: args[len++] = MDOC_Literal;
614: args[len++] = MDOC_File;
615: args[len++] = MDOC_Offset;
616: args[len++] = MDOC_Compact;
617: args[len++] = MDOC_Centred;
1.1 kristaps 618: break;
619: case (MDOC_Bf):
1.63 kristaps 620: args[len++] = MDOC_Emphasis;
621: args[len++] = MDOC_Literal;
622: args[len++] = MDOC_Symbolic;
1.1 kristaps 623: break;
624: case (MDOC_Bk):
1.63 kristaps 625: args[len++] = MDOC_Words;
1.1 kristaps 626: break;
627: case (MDOC_Bl):
1.63 kristaps 628: args[len++] = MDOC_Bullet;
629: args[len++] = MDOC_Dash;
630: args[len++] = MDOC_Hyphen;
631: args[len++] = MDOC_Item;
632: args[len++] = MDOC_Enum;
633: args[len++] = MDOC_Tag;
634: args[len++] = MDOC_Diag;
635: args[len++] = MDOC_Hang;
636: args[len++] = MDOC_Ohang;
637: args[len++] = MDOC_Inset;
638: args[len++] = MDOC_Column;
639: args[len++] = MDOC_Width;
640: args[len++] = MDOC_Offset;
641: args[len++] = MDOC_Compact;
642: args[len++] = MDOC_Nested;
1.1 kristaps 643: break;
644: case (MDOC_Rv):
645: /* FALLTHROUGH */
646: case (MDOC_Ex):
1.63 kristaps 647: args[len++] = MDOC_Std;
1.1 kristaps 648: break;
649: default:
650: break;
651: }
1.63 kristaps 652:
653: for (i = 0; i < len; i++)
654: if (0 == strcmp(p, mdoc_argnames[args[i]]))
655: return(args[i]);
1.1 kristaps 656:
657: return(MDOC_ARG_MAX);
658: }
659:
660: static int
1.11 kristaps 661: argv_multi(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: for (v->sz = 0; ; v->sz++) {
668: if ('-' == buf[*pos])
669: break;
1.43 kristaps 670: ac = args(m, line, pos, buf, 0, &p);
671: if (ARGS_ERROR == ac)
1.1 kristaps 672: return(0);
1.43 kristaps 673: else if (ARGS_EOLN == ac)
1.1 kristaps 674: break;
675:
1.31 kristaps 676: if (0 == v->sz % MULTI_STEP)
677: v->value = mandoc_realloc(v->value,
1.1 kristaps 678: (v->sz + MULTI_STEP) * sizeof(char *));
1.31 kristaps 679:
680: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 681: }
682:
1.7 kristaps 683: return(1);
1.1 kristaps 684: }
685:
686: static int
1.11 kristaps 687: argv_opt_single(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: if ('-' == buf[*pos])
694: return(1);
695:
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: if (ARGS_EOLN == ac)
1.1 kristaps 700: return(1);
701:
702: v->sz = 1;
1.31 kristaps 703: v->value = mandoc_malloc(sizeof(char *));
704: v->value[0] = mandoc_strdup(p);
1.2 kristaps 705:
1.1 kristaps 706: return(1);
707: }
708:
709: /*
710: * Parse a single, mandatory value from the stream.
711: */
712: static int
1.11 kristaps 713: argv_single(struct mdoc *m, int line,
1.1 kristaps 714: struct mdoc_argv *v, int *pos, char *buf)
715: {
1.43 kristaps 716: int ppos;
717: enum margserr ac;
1.1 kristaps 718: char *p;
719:
720: ppos = *pos;
721:
1.43 kristaps 722: ac = args(m, line, pos, buf, 0, &p);
1.49 kristaps 723: if (ARGS_EOLN == ac) {
724: mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTARGVCOUNT);
725: return(0);
726: } else if (ARGS_ERROR == ac)
1.1 kristaps 727: return(0);
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: * Determine rules for parsing arguments. Arguments can either accept
738: * no parameters, an optional single parameter, one parameter, or
739: * multiple parameters.
740: */
741: static int
742: argv(struct mdoc *mdoc, int line,
743: struct mdoc_argv *v, int *pos, char *buf)
744: {
745:
746: v->sz = 0;
747: v->value = NULL;
748:
1.68 kristaps 749: switch (argvflags[v->arg]) {
1.1 kristaps 750: case (ARGV_SINGLE):
751: return(argv_single(mdoc, line, v, pos, buf));
752: case (ARGV_MULTI):
753: return(argv_multi(mdoc, line, v, pos, buf));
754: case (ARGV_OPT_SINGLE):
755: return(argv_opt_single(mdoc, line, v, pos, buf));
1.68 kristaps 756: case (ARGV_NONE):
757: break;
1.1 kristaps 758: default:
1.68 kristaps 759: abort();
760: /* NOTREACHED */
1.1 kristaps 761: }
762:
763: return(1);
764: }
CVSweb