Annotation of mandoc/mdoc_argv.c, Revision 1.68
1.68 ! kristaps 1: /* $Id: mdoc_argv.c,v 1.67 2011/03/17 10:41:18 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 **);
1.64 kristaps 38: static int args_checkpunct(const char *);
1.1 kristaps 39: static int argv(struct mdoc *, int,
40: struct mdoc_argv *, int *, char *);
41: static int argv_single(struct mdoc *, int,
42: struct mdoc_argv *, int *, char *);
43: static int argv_opt_single(struct mdoc *, int,
44: struct mdoc_argv *, int *, char *);
45: static int argv_multi(struct mdoc *, int,
46: struct mdoc_argv *, int *, char *);
1.2 kristaps 47:
1.68 ! kristaps 48: enum argvflag {
! 49: ARGV_NONE, /* no args to flag (e.g., -split) */
! 50: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
! 51: ARGV_MULTI, /* multiple args (e.g., -column xxx yyy) */
! 52: ARGV_OPT_SINGLE /* optional arg (e.g., -offset [xxx]) */
! 53: };
1.1 kristaps 54:
1.68 ! kristaps 55: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 56: ARGV_NONE, /* MDOC_Split */
57: ARGV_NONE, /* MDOC_Nosplit */
58: ARGV_NONE, /* MDOC_Ragged */
59: ARGV_NONE, /* MDOC_Unfilled */
60: ARGV_NONE, /* MDOC_Literal */
1.29 kristaps 61: ARGV_SINGLE, /* MDOC_File */
1.26 kristaps 62: ARGV_OPT_SINGLE, /* MDOC_Offset */
1.1 kristaps 63: ARGV_NONE, /* MDOC_Bullet */
64: ARGV_NONE, /* MDOC_Dash */
65: ARGV_NONE, /* MDOC_Hyphen */
66: ARGV_NONE, /* MDOC_Item */
67: ARGV_NONE, /* MDOC_Enum */
68: ARGV_NONE, /* MDOC_Tag */
69: ARGV_NONE, /* MDOC_Diag */
70: ARGV_NONE, /* MDOC_Hang */
71: ARGV_NONE, /* MDOC_Ohang */
72: ARGV_NONE, /* MDOC_Inset */
73: ARGV_MULTI, /* MDOC_Column */
74: ARGV_SINGLE, /* MDOC_Width */
75: ARGV_NONE, /* MDOC_Compact */
1.25 kristaps 76: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 77: ARGV_NONE, /* MDOC_Filled */
78: ARGV_NONE, /* MDOC_Words */
79: ARGV_NONE, /* MDOC_Emphasis */
80: ARGV_NONE, /* MDOC_Symbolic */
81: ARGV_NONE /* MDOC_Symbolic */
82: };
83:
1.68 ! kristaps 84: static const int argflags[MDOC_MAX] = {
1.5 kristaps 85: 0, /* Ap */
1.1 kristaps 86: 0, /* Dd */
87: 0, /* Dt */
88: 0, /* Os */
1.17 kristaps 89: 0, /* Sh */
90: 0, /* Ss */
1.59 kristaps 91: 0, /* Pp */
1.1 kristaps 92: ARGS_DELIM, /* D1 */
1.17 kristaps 93: ARGS_DELIM, /* Dl */
1.1 kristaps 94: 0, /* Bd */
95: 0, /* Ed */
1.17 kristaps 96: 0, /* Bl */
1.1 kristaps 97: 0, /* El */
98: 0, /* It */
99: ARGS_DELIM, /* Ad */
100: ARGS_DELIM, /* An */
1.17 kristaps 101: ARGS_DELIM, /* Ar */
102: 0, /* Cd */
1.1 kristaps 103: ARGS_DELIM, /* Cm */
104: ARGS_DELIM, /* Dv */
105: ARGS_DELIM, /* Er */
106: ARGS_DELIM, /* Ev */
107: 0, /* Ex */
1.17 kristaps 108: ARGS_DELIM, /* Fa */
1.1 kristaps 109: 0, /* Fd */
110: ARGS_DELIM, /* Fl */
1.17 kristaps 111: ARGS_DELIM, /* Fn */
112: ARGS_DELIM, /* Ft */
1.1 kristaps 113: ARGS_DELIM, /* Ic */
114: 0, /* In */
1.17 kristaps 115: ARGS_DELIM, /* Li */
116: 0, /* Nd */
1.1 kristaps 117: ARGS_DELIM, /* Nm */
118: ARGS_DELIM, /* Op */
119: 0, /* Ot */
120: ARGS_DELIM, /* Pa */
121: 0, /* Rv */
1.15 kristaps 122: ARGS_DELIM, /* St */
1.1 kristaps 123: ARGS_DELIM, /* Va */
124: ARGS_DELIM, /* Vt */
125: ARGS_DELIM, /* Xr */
1.17 kristaps 126: 0, /* %A */
127: 0, /* %B */
128: 0, /* %D */
129: 0, /* %I */
130: 0, /* %J */
131: 0, /* %N */
132: 0, /* %O */
133: 0, /* %P */
134: 0, /* %R */
135: 0, /* %T */
136: 0, /* %V */
1.1 kristaps 137: ARGS_DELIM, /* Ac */
138: 0, /* Ao */
139: ARGS_DELIM, /* Aq */
140: ARGS_DELIM, /* At */
141: ARGS_DELIM, /* Bc */
142: 0, /* Bf */
143: 0, /* Bo */
144: ARGS_DELIM, /* Bq */
145: ARGS_DELIM, /* Bsx */
146: ARGS_DELIM, /* Bx */
147: 0, /* Db */
148: ARGS_DELIM, /* Dc */
149: 0, /* Do */
150: ARGS_DELIM, /* Dq */
151: ARGS_DELIM, /* Ec */
152: 0, /* Ef */
153: ARGS_DELIM, /* Em */
154: 0, /* Eo */
155: ARGS_DELIM, /* Fx */
156: ARGS_DELIM, /* Ms */
157: ARGS_DELIM, /* No */
158: ARGS_DELIM, /* Ns */
159: ARGS_DELIM, /* Nx */
160: ARGS_DELIM, /* Ox */
161: ARGS_DELIM, /* Pc */
162: ARGS_DELIM, /* Pf */
163: 0, /* Po */
164: ARGS_DELIM, /* Pq */
165: ARGS_DELIM, /* Qc */
166: ARGS_DELIM, /* Ql */
167: 0, /* Qo */
168: ARGS_DELIM, /* Qq */
169: 0, /* Re */
170: 0, /* Rs */
171: ARGS_DELIM, /* Sc */
172: 0, /* So */
173: ARGS_DELIM, /* Sq */
174: 0, /* Sm */
175: ARGS_DELIM, /* Sx */
1.17 kristaps 176: ARGS_DELIM, /* Sy */
1.1 kristaps 177: ARGS_DELIM, /* Tn */
178: ARGS_DELIM, /* Ux */
179: ARGS_DELIM, /* Xc */
180: 0, /* Xo */
1.17 kristaps 181: 0, /* Fo */
1.1 kristaps 182: 0, /* Fc */
183: 0, /* Oo */
184: ARGS_DELIM, /* Oc */
185: 0, /* Bk */
186: 0, /* Ek */
187: 0, /* Bt */
188: 0, /* Hf */
189: 0, /* Fr */
190: 0, /* Ud */
191: 0, /* Lb */
1.60 schwarze 192: 0, /* Lp */
1.17 kristaps 193: ARGS_DELIM, /* Lk */
194: ARGS_DELIM, /* Mt */
1.1 kristaps 195: ARGS_DELIM, /* Brq */
196: 0, /* Bro */
197: ARGS_DELIM, /* Brc */
1.17 kristaps 198: 0, /* %C */
1.1 kristaps 199: 0, /* Es */
200: 0, /* En */
201: 0, /* Dx */
1.17 kristaps 202: 0, /* %Q */
1.14 kristaps 203: 0, /* br */
204: 0, /* sp */
1.30 kristaps 205: 0, /* %U */
1.51 kristaps 206: 0, /* Ta */
1.1 kristaps 207: };
208:
209: /*
210: * Parse an argument from line text. This comes in the form of -key
211: * [value0...], which may either have a single mandatory value, at least
212: * one mandatory value, an optional single value, or no value.
213: */
1.40 kristaps 214: enum margverr
1.35 kristaps 215: mdoc_argv(struct mdoc *m, int line, enum mdoct tok,
1.1 kristaps 216: struct mdoc_arg **v, int *pos, char *buf)
217: {
218: char *p, sv;
219: struct mdoc_argv tmp;
220: struct mdoc_arg *arg;
221:
1.52 kristaps 222: if ('\0' == buf[*pos])
1.1 kristaps 223: return(ARGV_EOLN);
224:
225: assert(' ' != buf[*pos]);
226:
227: /* Parse through to the first unescaped space. */
228:
229: p = &buf[++(*pos)];
230:
231: assert(*pos > 0);
232:
233: /* LINTED */
234: while (buf[*pos]) {
235: if (' ' == buf[*pos])
236: if ('\\' != buf[*pos - 1])
237: break;
238: (*pos)++;
239: }
240:
241: /* XXX - save zeroed byte, if not an argument. */
242:
1.52 kristaps 243: sv = '\0';
1.1 kristaps 244: if (buf[*pos]) {
245: sv = buf[*pos];
1.52 kristaps 246: buf[(*pos)++] = '\0';
1.1 kristaps 247: }
248:
1.68 ! kristaps 249: memset(&tmp, 0, sizeof(struct mdoc_argv));
1.1 kristaps 250: tmp.line = line;
251: tmp.pos = *pos;
252:
253: /* See if our token accepts the argument. */
254:
255: if (MDOC_ARG_MAX == (tmp.arg = argv_a2arg(tok, p))) {
256: /* XXX - restore saved zeroed byte. */
257: if (sv)
258: buf[*pos - 1] = sv;
259: return(ARGV_WORD);
260: }
261:
262: while (buf[*pos] && ' ' == buf[*pos])
263: (*pos)++;
264:
1.11 kristaps 265: if ( ! argv(m, line, &tmp, pos, buf))
1.1 kristaps 266: return(ARGV_ERROR);
267:
1.31 kristaps 268: if (NULL == (arg = *v))
1.32 kristaps 269: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 270:
271: arg->argc++;
1.31 kristaps 272: arg->argv = mandoc_realloc
273: (arg->argv, arg->argc * sizeof(struct mdoc_argv));
1.2 kristaps 274:
1.68 ! kristaps 275: memcpy(&arg->argv[(int)arg->argc - 1],
1.1 kristaps 276: &tmp, sizeof(struct mdoc_argv));
277:
278: return(ARGV_ARG);
279: }
280:
281: void
282: mdoc_argv_free(struct mdoc_arg *p)
283: {
1.37 kristaps 284: int i;
1.1 kristaps 285:
286: if (NULL == p)
287: return;
288:
289: if (p->refcnt) {
290: --(p->refcnt);
291: if (p->refcnt)
292: return;
293: }
294: assert(p->argc);
295:
1.37 kristaps 296: for (i = (int)p->argc - 1; i >= 0; i--)
297: mdoc_argn_free(p, i);
298:
299: free(p->argv);
300: free(p);
301: }
302:
303: void
304: mdoc_argn_free(struct mdoc_arg *p, int iarg)
305: {
1.58 kristaps 306: struct mdoc_argv *arg;
1.37 kristaps 307: int j;
1.58 kristaps 308:
309: arg = &p->argv[iarg];
1.1 kristaps 310:
1.37 kristaps 311: if (arg->sz && arg->value) {
312: for (j = (int)arg->sz - 1; j >= 0; j--)
313: free(arg->value[j]);
314: free(arg->value);
1.1 kristaps 315: }
316:
1.37 kristaps 317: for (--p->argc; iarg < (int)p->argc; iarg++)
318: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 319: }
320:
1.39 kristaps 321: enum margserr
1.23 kristaps 322: mdoc_zargs(struct mdoc *m, int line, int *pos,
323: char *buf, int flags, char **v)
1.19 kristaps 324: {
325:
1.23 kristaps 326: return(args(m, line, pos, buf, flags, v));
1.19 kristaps 327: }
328:
1.39 kristaps 329: enum margserr
1.35 kristaps 330: mdoc_args(struct mdoc *m, int line, int *pos,
331: char *buf, enum mdoct tok, char **v)
1.1 kristaps 332: {
1.50 kristaps 333: int fl;
1.1 kristaps 334: struct mdoc_node *n;
335:
1.68 ! kristaps 336: fl = argflags[tok];
1.1 kristaps 337:
1.17 kristaps 338: if (MDOC_It != tok)
339: return(args(m, line, pos, buf, fl, v));
340:
1.51 kristaps 341: /*
342: * We know that we're in an `It', so it's reasonable to expect
343: * us to be sitting in a `Bl'. Someday this may not be the case
344: * (if we allow random `It's sitting out there), so provide a
345: * safe fall-back into the default behaviour.
346: */
347:
1.17 kristaps 348: for (n = m->last; n; n = n->parent)
1.50 kristaps 349: if (MDOC_Bl == n->tok)
1.17 kristaps 350: break;
1.1 kristaps 351:
1.62 kristaps 352: if (n && LIST_column == n->norm->Bl.type) {
1.17 kristaps 353: fl |= ARGS_TABSEP;
354: fl &= ~ARGS_DELIM;
1.1 kristaps 355: }
356:
1.11 kristaps 357: return(args(m, line, pos, buf, fl, v));
1.1 kristaps 358: }
359:
1.39 kristaps 360: static enum margserr
1.13 kristaps 361: args(struct mdoc *m, int line, int *pos,
362: char *buf, int fl, char **v)
1.1 kristaps 363: {
1.52 kristaps 364: int i;
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:
1.64 kristaps 402: if (ARGS_DELIM & fl && args_checkpunct(&buf[*pos])) {
403: i = strlen(&buf[*pos]) + *pos;
404: if (i && ' ' != buf[i - 1])
405: return(ARGS_PUNCT);
406: if (ARGS_NOWARN & fl)
1.1 kristaps 407: return(ARGS_PUNCT);
1.67 kristaps 408: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.64 kristaps 409: return(ARGS_PUNCT);
1.1 kristaps 410: }
411:
1.17 kristaps 412: /*
413: * First handle TABSEP items, restricted to `Bl -column'. This
414: * ignores conventional token parsing and instead uses tabs or
415: * `Ta' macros to separate phrases. Phrases are parsed again
416: * for arguments at a later phase.
417: */
1.1 kristaps 418:
1.17 kristaps 419: if (ARGS_TABSEP & fl) {
1.20 kristaps 420: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 421: p = strchr(*v, '\t');
1.46 kristaps 422: pp = NULL;
1.17 kristaps 423:
424: /* Scan ahead to unescaped `Ta'. */
1.46 kristaps 425: if ( ! (MDOC_PHRASELIT & m->flags))
426: for (pp = *v; ; pp++) {
427: if (NULL == (pp = strstr(pp, "Ta")))
428: break;
429: if (pp > *v && ' ' != *(pp - 1))
430: continue;
1.51 kristaps 431: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 432: break;
433: }
1.17 kristaps 434:
1.42 kristaps 435: /* By default, assume a phrase. */
436: rc = ARGS_PHRASE;
437:
1.21 kristaps 438: /*
439: * Adjust new-buffer position to be beyond delimiter
440: * mark (e.g., Ta -> end + 2).
441: */
1.20 kristaps 442: if (p && pp) {
443: *pos += pp < p ? 2 : 1;
1.42 kristaps 444: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 445: p = pp < p ? pp : p;
446: } else if (p && ! pp) {
1.42 kristaps 447: rc = ARGS_PPHRASE;
1.20 kristaps 448: *pos += 1;
449: } else if (pp && ! p) {
1.17 kristaps 450: p = pp;
1.20 kristaps 451: *pos += 2;
1.44 kristaps 452: } else {
453: rc = ARGS_PEND;
1.20 kristaps 454: p = strchr(*v, 0);
1.44 kristaps 455: }
1.20 kristaps 456:
1.21 kristaps 457: /* Whitespace check for eoln case... */
1.52 kristaps 458: if ('\0' == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl))
1.67 kristaps 459: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.20 kristaps 460:
461: *pos += (int)(p - *v);
1.17 kristaps 462:
463: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 464: pp = p - 1;
465: while (pp > *v && ' ' == *pp) {
466: if (pp > *v && '\\' == *(pp - 1))
467: break;
468: pp--;
1.17 kristaps 469: }
1.20 kristaps 470: *(pp + 1) = 0;
1.17 kristaps 471:
1.20 kristaps 472: /* Strip delimiter's proceeding whitespace. */
473: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
474: /* Skip ahead. */ ;
1.1 kristaps 475:
1.41 kristaps 476: return(rc);
1.17 kristaps 477: }
1.1 kristaps 478:
1.17 kristaps 479: /*
480: * Process a quoted literal. A quote begins with a double-quote
481: * and ends with a double-quote NOT preceded by a double-quote.
482: * Whitespace is NOT involved in literal termination.
483: */
1.1 kristaps 484:
1.46 kristaps 485: if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
486: if ( ! (MDOC_PHRASELIT & m->flags))
487: *v = &buf[++(*pos)];
488:
1.47 kristaps 489: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 490: m->flags |= MDOC_PHRASELIT;
1.1 kristaps 491:
1.17 kristaps 492: for ( ; buf[*pos]; (*pos)++) {
493: if ('\"' != buf[*pos])
494: continue;
495: if ('\"' != buf[*pos + 1])
496: break;
497: (*pos)++;
498: }
1.1 kristaps 499:
1.46 kristaps 500: if ('\0' == buf[*pos]) {
1.47 kristaps 501: if (ARGS_NOWARN & fl || MDOC_PPHRASE & m->flags)
1.23 kristaps 502: return(ARGS_QWORD);
1.67 kristaps 503: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.18 kristaps 504: return(ARGS_QWORD);
1.17 kristaps 505: }
1.1 kristaps 506:
1.46 kristaps 507: m->flags &= ~MDOC_PHRASELIT;
508: buf[(*pos)++] = '\0';
1.1 kristaps 509:
1.46 kristaps 510: if ('\0' == buf[*pos])
1.17 kristaps 511: return(ARGS_QWORD);
1.1 kristaps 512:
1.17 kristaps 513: while (' ' == buf[*pos])
514: (*pos)++;
1.1 kristaps 515:
1.23 kristaps 516: if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
1.67 kristaps 517: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 518:
1.17 kristaps 519: return(ARGS_QWORD);
1.1 kristaps 520: }
521:
1.17 kristaps 522: /*
523: * A non-quoted term progresses until either the end of line or
524: * a non-escaped whitespace.
1.1 kristaps 525: */
526:
1.17 kristaps 527: for ( ; buf[*pos]; (*pos)++)
1.52 kristaps 528: if (*pos && ' ' == buf[*pos] && '\\' != buf[*pos - 1])
1.17 kristaps 529: break;
1.1 kristaps 530:
1.46 kristaps 531: if ('\0' == buf[*pos])
1.17 kristaps 532: return(ARGS_WORD);
1.1 kristaps 533:
1.46 kristaps 534: buf[(*pos)++] = '\0';
1.1 kristaps 535:
1.17 kristaps 536: while (' ' == buf[*pos])
1.1 kristaps 537: (*pos)++;
538:
1.46 kristaps 539: if ('\0' == buf[*pos] && ! (ARGS_NOWARN & fl))
1.67 kristaps 540: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 541:
1.17 kristaps 542: return(ARGS_WORD);
1.64 kristaps 543: }
544:
545: /*
546: * Check if the string consists only of space-separated closing
1.65 kristaps 547: * delimiters. This is a bit of a dance: the first must be a close
548: * delimiter, but it may be followed by middle delimiters. Arbitrary
549: * whitespace may separate these tokens.
1.64 kristaps 550: */
551: static int
552: args_checkpunct(const char *p)
553: {
1.65 kristaps 554: int i, j;
555: char buf[DELIMSZ];
1.64 kristaps 556: enum mdelim d;
557:
558: i = 0;
559:
1.65 kristaps 560: /* First token must be a close-delimiter. */
561:
562: for (j = 0; p[i] && ' ' != p[i] && j < DELIMSZ; j++, i++)
563: buf[j] = p[i];
564:
565: if (DELIMSZ == j)
566: return(0);
567:
568: buf[j] = '\0';
1.66 kristaps 569: if (DELIM_CLOSE != mandoc_isdelim(buf))
1.64 kristaps 570: return(0);
571:
1.65 kristaps 572: while (' ' == p[i])
573: i++;
574:
575: /* Remaining must NOT be open/none. */
576:
577: while (p[i]) {
578: j = 0;
579: while (p[i] && ' ' != p[i] && j < DELIMSZ)
580: buf[j++] = p[i++];
581:
582: if (DELIMSZ == j)
583: return(0);
584:
585: buf[j] = '\0';
1.66 kristaps 586: d = mandoc_isdelim(buf);
1.64 kristaps 587: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 588: return(0);
589:
590: while (' ' == p[i])
1.64 kristaps 591: i++;
592: }
593:
594: return('\0' == p[i]);
1.1 kristaps 595: }
596:
1.63 kristaps 597: /*
598: * Match up an argument string (e.g., `-foo bar' having "foo") with the
599: * correrct identifier. It must apply to the given macro. If none was
600: * found (including bad matches), return MDOC_ARG_MAX.
601: */
1.57 kristaps 602: static enum mdocargt
1.35 kristaps 603: argv_a2arg(enum mdoct tok, const char *p)
1.1 kristaps 604: {
1.63 kristaps 605: enum mdocargt args[MDOC_ARG_MAX];
606: int i, len;
1.1 kristaps 607:
1.63 kristaps 608: len = 0;
1.1 kristaps 609:
610: switch (tok) {
611: case (MDOC_An):
1.63 kristaps 612: args[len++] = MDOC_Split;
613: args[len++] = MDOC_Nosplit;
1.1 kristaps 614: break;
615: case (MDOC_Bd):
1.63 kristaps 616: args[len++] = MDOC_Ragged;
617: args[len++] = MDOC_Unfilled;
618: args[len++] = MDOC_Filled;
619: args[len++] = MDOC_Literal;
620: args[len++] = MDOC_File;
621: args[len++] = MDOC_Offset;
622: args[len++] = MDOC_Compact;
623: args[len++] = MDOC_Centred;
1.1 kristaps 624: break;
625: case (MDOC_Bf):
1.63 kristaps 626: args[len++] = MDOC_Emphasis;
627: args[len++] = MDOC_Literal;
628: args[len++] = MDOC_Symbolic;
1.1 kristaps 629: break;
630: case (MDOC_Bk):
1.63 kristaps 631: args[len++] = MDOC_Words;
1.1 kristaps 632: break;
633: case (MDOC_Bl):
1.63 kristaps 634: args[len++] = MDOC_Bullet;
635: args[len++] = MDOC_Dash;
636: args[len++] = MDOC_Hyphen;
637: args[len++] = MDOC_Item;
638: args[len++] = MDOC_Enum;
639: args[len++] = MDOC_Tag;
640: args[len++] = MDOC_Diag;
641: args[len++] = MDOC_Hang;
642: args[len++] = MDOC_Ohang;
643: args[len++] = MDOC_Inset;
644: args[len++] = MDOC_Column;
645: args[len++] = MDOC_Width;
646: args[len++] = MDOC_Offset;
647: args[len++] = MDOC_Compact;
648: args[len++] = MDOC_Nested;
1.1 kristaps 649: break;
650: case (MDOC_Rv):
651: /* FALLTHROUGH */
652: case (MDOC_Ex):
1.63 kristaps 653: args[len++] = MDOC_Std;
1.1 kristaps 654: break;
655: default:
656: break;
657: }
1.63 kristaps 658:
659: for (i = 0; i < len; i++)
660: if (0 == strcmp(p, mdoc_argnames[args[i]]))
661: return(args[i]);
1.1 kristaps 662:
663: return(MDOC_ARG_MAX);
664: }
665:
666: static int
1.11 kristaps 667: argv_multi(struct mdoc *m, int line,
1.1 kristaps 668: struct mdoc_argv *v, int *pos, char *buf)
669: {
1.43 kristaps 670: enum margserr ac;
1.1 kristaps 671: char *p;
672:
673: for (v->sz = 0; ; v->sz++) {
674: if ('-' == buf[*pos])
675: break;
1.43 kristaps 676: ac = args(m, line, pos, buf, 0, &p);
677: if (ARGS_ERROR == ac)
1.1 kristaps 678: return(0);
1.43 kristaps 679: else if (ARGS_EOLN == ac)
1.1 kristaps 680: break;
681:
1.31 kristaps 682: if (0 == v->sz % MULTI_STEP)
683: v->value = mandoc_realloc(v->value,
1.1 kristaps 684: (v->sz + MULTI_STEP) * sizeof(char *));
1.31 kristaps 685:
686: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 687: }
688:
1.7 kristaps 689: return(1);
1.1 kristaps 690: }
691:
692: static int
1.11 kristaps 693: argv_opt_single(struct mdoc *m, int line,
1.1 kristaps 694: struct mdoc_argv *v, int *pos, char *buf)
695: {
1.43 kristaps 696: enum margserr ac;
1.1 kristaps 697: char *p;
698:
699: if ('-' == buf[*pos])
700: return(1);
701:
1.43 kristaps 702: ac = args(m, line, pos, buf, 0, &p);
703: if (ARGS_ERROR == ac)
1.1 kristaps 704: return(0);
1.43 kristaps 705: if (ARGS_EOLN == ac)
1.1 kristaps 706: return(1);
707:
708: v->sz = 1;
1.31 kristaps 709: v->value = mandoc_malloc(sizeof(char *));
710: v->value[0] = mandoc_strdup(p);
1.2 kristaps 711:
1.1 kristaps 712: return(1);
713: }
714:
715: /*
716: * Parse a single, mandatory value from the stream.
717: */
718: static int
1.11 kristaps 719: argv_single(struct mdoc *m, int line,
1.1 kristaps 720: struct mdoc_argv *v, int *pos, char *buf)
721: {
1.43 kristaps 722: int ppos;
723: enum margserr ac;
1.1 kristaps 724: char *p;
725:
726: ppos = *pos;
727:
1.43 kristaps 728: ac = args(m, line, pos, buf, 0, &p);
1.49 kristaps 729: if (ARGS_EOLN == ac) {
730: mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTARGVCOUNT);
731: return(0);
732: } else if (ARGS_ERROR == ac)
1.1 kristaps 733: return(0);
734:
735: v->sz = 1;
1.31 kristaps 736: v->value = mandoc_malloc(sizeof(char *));
737: v->value[0] = mandoc_strdup(p);
1.2 kristaps 738:
1.1 kristaps 739: return(1);
740: }
741:
742: /*
743: * Determine rules for parsing arguments. Arguments can either accept
744: * no parameters, an optional single parameter, one parameter, or
745: * multiple parameters.
746: */
747: static int
748: argv(struct mdoc *mdoc, int line,
749: struct mdoc_argv *v, int *pos, char *buf)
750: {
751:
752: v->sz = 0;
753: v->value = NULL;
754:
1.68 ! kristaps 755: switch (argvflags[v->arg]) {
1.1 kristaps 756: case (ARGV_SINGLE):
757: return(argv_single(mdoc, line, v, pos, buf));
758: case (ARGV_MULTI):
759: return(argv_multi(mdoc, line, v, pos, buf));
760: case (ARGV_OPT_SINGLE):
761: return(argv_opt_single(mdoc, line, v, pos, buf));
1.68 ! kristaps 762: case (ARGV_NONE):
! 763: break;
1.1 kristaps 764: default:
1.68 ! kristaps 765: abort();
! 766: /* NOTREACHED */
1.1 kristaps 767: }
768:
769: return(1);
770: }
CVSweb