Annotation of mandoc/mdoc_argv.c, Revision 1.77
1.77 ! kristaps 1: /* $Id: mdoc_argv.c,v 1.76 2011/04/19 16:38:48 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:
1.72 kristaps 29: #include "mdoc.h"
1.49 kristaps 30: #include "mandoc.h"
1.1 kristaps 31: #include "libmdoc.h"
1.31 kristaps 32: #include "libmandoc.h"
1.1 kristaps 33:
1.68 kristaps 34: #define MULTI_STEP 5 /* pre-allocate argument values */
1.76 kristaps 35: #define DELIMSZ 6 /* max possible size of a delimiter */
36:
37: enum argsflag {
38: ARGSFL_NONE = 0,
39: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
40: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
41: };
42:
43: enum argvflag {
44: ARGV_NONE, /* no args to flag (e.g., -split) */
45: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
46: ARGV_MULTI, /* multiple args (e.g., -column xxx yyy) */
47: ARGV_OPT_SINGLE /* optional arg (e.g., -offset [xxx]) */
48: };
1.1 kristaps 49:
1.57 kristaps 50: static enum mdocargt argv_a2arg(enum mdoct, const char *);
1.39 kristaps 51: static enum margserr args(struct mdoc *, int, int *,
1.76 kristaps 52: char *, enum argsflag, char **);
1.77 ! kristaps 53: static int args_checkpunct(const char *, int);
1.1 kristaps 54: static int argv(struct mdoc *, int,
55: struct mdoc_argv *, int *, char *);
56: static int argv_single(struct mdoc *, int,
57: struct mdoc_argv *, int *, char *);
58: static int argv_opt_single(struct mdoc *, int,
59: struct mdoc_argv *, int *, char *);
60: static int argv_multi(struct mdoc *, int,
61: struct mdoc_argv *, int *, char *);
1.71 kristaps 62: static void argn_free(struct mdoc_arg *, int);
1.2 kristaps 63:
1.68 kristaps 64: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 65: ARGV_NONE, /* MDOC_Split */
66: ARGV_NONE, /* MDOC_Nosplit */
67: ARGV_NONE, /* MDOC_Ragged */
68: ARGV_NONE, /* MDOC_Unfilled */
69: ARGV_NONE, /* MDOC_Literal */
1.29 kristaps 70: ARGV_SINGLE, /* MDOC_File */
1.26 kristaps 71: ARGV_OPT_SINGLE, /* MDOC_Offset */
1.1 kristaps 72: ARGV_NONE, /* MDOC_Bullet */
73: ARGV_NONE, /* MDOC_Dash */
74: ARGV_NONE, /* MDOC_Hyphen */
75: ARGV_NONE, /* MDOC_Item */
76: ARGV_NONE, /* MDOC_Enum */
77: ARGV_NONE, /* MDOC_Tag */
78: ARGV_NONE, /* MDOC_Diag */
79: ARGV_NONE, /* MDOC_Hang */
80: ARGV_NONE, /* MDOC_Ohang */
81: ARGV_NONE, /* MDOC_Inset */
82: ARGV_MULTI, /* MDOC_Column */
83: ARGV_SINGLE, /* MDOC_Width */
84: ARGV_NONE, /* MDOC_Compact */
1.25 kristaps 85: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 86: ARGV_NONE, /* MDOC_Filled */
87: ARGV_NONE, /* MDOC_Words */
88: ARGV_NONE, /* MDOC_Emphasis */
89: ARGV_NONE, /* MDOC_Symbolic */
90: ARGV_NONE /* MDOC_Symbolic */
91: };
92:
1.76 kristaps 93: static const enum argsflag argflags[MDOC_MAX] = {
94: ARGSFL_NONE, /* Ap */
95: ARGSFL_NONE, /* Dd */
96: ARGSFL_NONE, /* Dt */
97: ARGSFL_NONE, /* Os */
98: ARGSFL_NONE, /* Sh */
99: ARGSFL_NONE, /* Ss */
100: ARGSFL_NONE, /* Pp */
101: ARGSFL_DELIM, /* D1 */
102: ARGSFL_DELIM, /* Dl */
103: ARGSFL_NONE, /* Bd */
104: ARGSFL_NONE, /* Ed */
105: ARGSFL_NONE, /* Bl */
106: ARGSFL_NONE, /* El */
107: ARGSFL_NONE, /* It */
108: ARGSFL_DELIM, /* Ad */
109: ARGSFL_DELIM, /* An */
110: ARGSFL_DELIM, /* Ar */
111: ARGSFL_NONE, /* Cd */
112: ARGSFL_DELIM, /* Cm */
113: ARGSFL_DELIM, /* Dv */
114: ARGSFL_DELIM, /* Er */
115: ARGSFL_DELIM, /* Ev */
116: ARGSFL_NONE, /* Ex */
117: ARGSFL_DELIM, /* Fa */
118: ARGSFL_NONE, /* Fd */
119: ARGSFL_DELIM, /* Fl */
120: ARGSFL_DELIM, /* Fn */
121: ARGSFL_DELIM, /* Ft */
122: ARGSFL_DELIM, /* Ic */
123: ARGSFL_NONE, /* In */
124: ARGSFL_DELIM, /* Li */
125: ARGSFL_NONE, /* Nd */
126: ARGSFL_DELIM, /* Nm */
127: ARGSFL_DELIM, /* Op */
128: ARGSFL_NONE, /* Ot */
129: ARGSFL_DELIM, /* Pa */
130: ARGSFL_NONE, /* Rv */
131: ARGSFL_DELIM, /* St */
132: ARGSFL_DELIM, /* Va */
133: ARGSFL_DELIM, /* Vt */
134: ARGSFL_DELIM, /* Xr */
135: ARGSFL_NONE, /* %A */
136: ARGSFL_NONE, /* %B */
137: ARGSFL_NONE, /* %D */
138: ARGSFL_NONE, /* %I */
139: ARGSFL_NONE, /* %J */
140: ARGSFL_NONE, /* %N */
141: ARGSFL_NONE, /* %O */
142: ARGSFL_NONE, /* %P */
143: ARGSFL_NONE, /* %R */
144: ARGSFL_NONE, /* %T */
145: ARGSFL_NONE, /* %V */
146: ARGSFL_DELIM, /* Ac */
147: ARGSFL_NONE, /* Ao */
148: ARGSFL_DELIM, /* Aq */
149: ARGSFL_DELIM, /* At */
150: ARGSFL_DELIM, /* Bc */
151: ARGSFL_NONE, /* Bf */
152: ARGSFL_NONE, /* Bo */
153: ARGSFL_DELIM, /* Bq */
154: ARGSFL_DELIM, /* Bsx */
155: ARGSFL_DELIM, /* Bx */
156: ARGSFL_NONE, /* Db */
157: ARGSFL_DELIM, /* Dc */
158: ARGSFL_NONE, /* Do */
159: ARGSFL_DELIM, /* Dq */
160: ARGSFL_DELIM, /* Ec */
161: ARGSFL_NONE, /* Ef */
162: ARGSFL_DELIM, /* Em */
163: ARGSFL_NONE, /* Eo */
164: ARGSFL_DELIM, /* Fx */
165: ARGSFL_DELIM, /* Ms */
166: ARGSFL_DELIM, /* No */
167: ARGSFL_DELIM, /* Ns */
168: ARGSFL_DELIM, /* Nx */
169: ARGSFL_DELIM, /* Ox */
170: ARGSFL_DELIM, /* Pc */
171: ARGSFL_DELIM, /* Pf */
172: ARGSFL_NONE, /* Po */
173: ARGSFL_DELIM, /* Pq */
174: ARGSFL_DELIM, /* Qc */
175: ARGSFL_DELIM, /* Ql */
176: ARGSFL_NONE, /* Qo */
177: ARGSFL_DELIM, /* Qq */
178: ARGSFL_NONE, /* Re */
179: ARGSFL_NONE, /* Rs */
180: ARGSFL_DELIM, /* Sc */
181: ARGSFL_NONE, /* So */
182: ARGSFL_DELIM, /* Sq */
183: ARGSFL_NONE, /* Sm */
184: ARGSFL_DELIM, /* Sx */
185: ARGSFL_DELIM, /* Sy */
186: ARGSFL_DELIM, /* Tn */
187: ARGSFL_DELIM, /* Ux */
188: ARGSFL_DELIM, /* Xc */
189: ARGSFL_NONE, /* Xo */
190: ARGSFL_NONE, /* Fo */
191: ARGSFL_NONE, /* Fc */
192: ARGSFL_NONE, /* Oo */
193: ARGSFL_DELIM, /* Oc */
194: ARGSFL_NONE, /* Bk */
195: ARGSFL_NONE, /* Ek */
196: ARGSFL_NONE, /* Bt */
197: ARGSFL_NONE, /* Hf */
198: ARGSFL_NONE, /* Fr */
199: ARGSFL_NONE, /* Ud */
200: ARGSFL_NONE, /* Lb */
201: ARGSFL_NONE, /* Lp */
202: ARGSFL_DELIM, /* Lk */
203: ARGSFL_DELIM, /* Mt */
204: ARGSFL_DELIM, /* Brq */
205: ARGSFL_NONE, /* Bro */
206: ARGSFL_DELIM, /* Brc */
207: ARGSFL_NONE, /* %C */
208: ARGSFL_NONE, /* Es */
209: ARGSFL_NONE, /* En */
210: ARGSFL_NONE, /* Dx */
211: ARGSFL_NONE, /* %Q */
212: ARGSFL_NONE, /* br */
213: ARGSFL_NONE, /* sp */
214: ARGSFL_NONE, /* %U */
215: ARGSFL_NONE, /* Ta */
1.1 kristaps 216: };
217:
1.73 kristaps 218: static const enum mdocargt args_Ex[] = {
219: MDOC_Std,
220: MDOC_ARG_MAX
221: };
222:
223: static const enum mdocargt args_An[] = {
224: MDOC_Split,
225: MDOC_Nosplit,
226: MDOC_ARG_MAX
227: };
228:
229: static const enum mdocargt args_Bd[] = {
230: MDOC_Ragged,
231: MDOC_Unfilled,
232: MDOC_Filled,
233: MDOC_Literal,
234: MDOC_File,
235: MDOC_Offset,
236: MDOC_Compact,
237: MDOC_Centred,
238: MDOC_ARG_MAX
239: };
240:
241: static const enum mdocargt args_Bf[] = {
242: MDOC_Emphasis,
243: MDOC_Literal,
244: MDOC_Symbolic,
245: MDOC_ARG_MAX
246: };
247:
248: static const enum mdocargt args_Bk[] = {
249: MDOC_Words,
250: MDOC_ARG_MAX
251: };
252:
253: static const enum mdocargt args_Bl[] = {
254: MDOC_Bullet,
255: MDOC_Dash,
256: MDOC_Hyphen,
257: MDOC_Item,
258: MDOC_Enum,
259: MDOC_Tag,
260: MDOC_Diag,
261: MDOC_Hang,
262: MDOC_Ohang,
263: MDOC_Inset,
264: MDOC_Column,
265: MDOC_Width,
266: MDOC_Offset,
267: MDOC_Compact,
268: MDOC_Nested,
269: MDOC_ARG_MAX
270: };
271:
1.1 kristaps 272: /*
273: * Parse an argument from line text. This comes in the form of -key
274: * [value0...], which may either have a single mandatory value, at least
275: * one mandatory value, an optional single value, or no value.
276: */
1.40 kristaps 277: enum margverr
1.35 kristaps 278: mdoc_argv(struct mdoc *m, int line, enum mdoct tok,
1.1 kristaps 279: struct mdoc_arg **v, int *pos, char *buf)
280: {
281: char *p, sv;
282: struct mdoc_argv tmp;
283: struct mdoc_arg *arg;
284:
1.52 kristaps 285: if ('\0' == buf[*pos])
1.1 kristaps 286: return(ARGV_EOLN);
287:
288: assert(' ' != buf[*pos]);
289:
290: /* Parse through to the first unescaped space. */
291:
292: p = &buf[++(*pos)];
293:
294: assert(*pos > 0);
295:
296: /* LINTED */
297: while (buf[*pos]) {
298: if (' ' == buf[*pos])
299: if ('\\' != buf[*pos - 1])
300: break;
301: (*pos)++;
302: }
303:
304: /* XXX - save zeroed byte, if not an argument. */
305:
1.52 kristaps 306: sv = '\0';
1.1 kristaps 307: if (buf[*pos]) {
308: sv = buf[*pos];
1.52 kristaps 309: buf[(*pos)++] = '\0';
1.1 kristaps 310: }
311:
1.68 kristaps 312: memset(&tmp, 0, sizeof(struct mdoc_argv));
1.1 kristaps 313: tmp.line = line;
314: tmp.pos = *pos;
315:
316: /* See if our token accepts the argument. */
317:
318: if (MDOC_ARG_MAX == (tmp.arg = argv_a2arg(tok, p))) {
319: /* XXX - restore saved zeroed byte. */
320: if (sv)
321: buf[*pos - 1] = sv;
322: return(ARGV_WORD);
323: }
324:
325: while (buf[*pos] && ' ' == buf[*pos])
326: (*pos)++;
327:
1.11 kristaps 328: if ( ! argv(m, line, &tmp, pos, buf))
1.1 kristaps 329: return(ARGV_ERROR);
330:
1.31 kristaps 331: if (NULL == (arg = *v))
1.32 kristaps 332: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 333:
334: arg->argc++;
1.31 kristaps 335: arg->argv = mandoc_realloc
336: (arg->argv, arg->argc * sizeof(struct mdoc_argv));
1.2 kristaps 337:
1.68 kristaps 338: memcpy(&arg->argv[(int)arg->argc - 1],
1.1 kristaps 339: &tmp, sizeof(struct mdoc_argv));
340:
341: return(ARGV_ARG);
342: }
343:
344: void
345: mdoc_argv_free(struct mdoc_arg *p)
346: {
1.37 kristaps 347: int i;
1.1 kristaps 348:
349: if (NULL == p)
350: return;
351:
352: if (p->refcnt) {
353: --(p->refcnt);
354: if (p->refcnt)
355: return;
356: }
357: assert(p->argc);
358:
1.37 kristaps 359: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 360: argn_free(p, i);
1.37 kristaps 361:
362: free(p->argv);
363: free(p);
364: }
365:
1.71 kristaps 366: static void
367: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 368: {
1.58 kristaps 369: struct mdoc_argv *arg;
1.37 kristaps 370: int j;
1.58 kristaps 371:
372: arg = &p->argv[iarg];
1.1 kristaps 373:
1.37 kristaps 374: if (arg->sz && arg->value) {
375: for (j = (int)arg->sz - 1; j >= 0; j--)
376: free(arg->value[j]);
377: free(arg->value);
1.1 kristaps 378: }
379:
1.37 kristaps 380: for (--p->argc; iarg < (int)p->argc; iarg++)
381: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 382: }
383:
1.39 kristaps 384: enum margserr
1.76 kristaps 385: mdoc_zargs(struct mdoc *m, int line, int *pos, char *buf, char **v)
1.19 kristaps 386: {
387:
1.76 kristaps 388: return(args(m, line, pos, buf, ARGSFL_NONE, v));
1.19 kristaps 389: }
390:
1.39 kristaps 391: enum margserr
1.35 kristaps 392: mdoc_args(struct mdoc *m, int line, int *pos,
393: char *buf, enum mdoct tok, char **v)
1.1 kristaps 394: {
1.76 kristaps 395: enum argsflag fl;
1.1 kristaps 396: struct mdoc_node *n;
397:
1.68 kristaps 398: fl = argflags[tok];
1.1 kristaps 399:
1.17 kristaps 400: if (MDOC_It != tok)
401: return(args(m, line, pos, buf, fl, v));
402:
1.51 kristaps 403: /*
404: * We know that we're in an `It', so it's reasonable to expect
405: * us to be sitting in a `Bl'. Someday this may not be the case
406: * (if we allow random `It's sitting out there), so provide a
407: * safe fall-back into the default behaviour.
408: */
409:
1.17 kristaps 410: for (n = m->last; n; n = n->parent)
1.50 kristaps 411: if (MDOC_Bl == n->tok)
1.76 kristaps 412: if (LIST_column == n->norm->Bl.type) {
413: fl = ARGSFL_TABSEP;
414: break;
415: }
1.1 kristaps 416:
1.11 kristaps 417: return(args(m, line, pos, buf, fl, v));
1.1 kristaps 418: }
419:
1.39 kristaps 420: static enum margserr
1.13 kristaps 421: args(struct mdoc *m, int line, int *pos,
1.76 kristaps 422: char *buf, enum argsflag fl, char **v)
1.1 kristaps 423: {
1.52 kristaps 424: char *p, *pp;
425: enum margserr rc;
1.1 kristaps 426:
1.17 kristaps 427: assert(' ' != buf[*pos]);
1.1 kristaps 428:
1.46 kristaps 429: if ('\0' == buf[*pos]) {
1.47 kristaps 430: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 431: return(ARGS_EOLN);
432: /*
433: * If we're not in a partial phrase and the flag for
434: * being a phrase literal is still set, the punctuation
435: * is unterminated.
436: */
437: if (MDOC_PHRASELIT & m->flags)
1.67 kristaps 438: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.46 kristaps 439:
440: m->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 441: return(ARGS_EOLN);
1.46 kristaps 442: }
1.1 kristaps 443:
1.64 kristaps 444: *v = &buf[*pos];
1.17 kristaps 445:
1.76 kristaps 446: if (ARGSFL_DELIM == fl)
1.77 ! kristaps 447: if (args_checkpunct(buf, *pos))
1.76 kristaps 448: return(ARGS_PUNCT);
1.1 kristaps 449:
1.17 kristaps 450: /*
451: * First handle TABSEP items, restricted to `Bl -column'. This
452: * ignores conventional token parsing and instead uses tabs or
453: * `Ta' macros to separate phrases. Phrases are parsed again
454: * for arguments at a later phase.
455: */
1.1 kristaps 456:
1.76 kristaps 457: if (ARGSFL_TABSEP == fl) {
1.20 kristaps 458: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 459: p = strchr(*v, '\t');
1.46 kristaps 460: pp = NULL;
1.17 kristaps 461:
462: /* Scan ahead to unescaped `Ta'. */
1.46 kristaps 463: if ( ! (MDOC_PHRASELIT & m->flags))
464: for (pp = *v; ; pp++) {
465: if (NULL == (pp = strstr(pp, "Ta")))
466: break;
467: if (pp > *v && ' ' != *(pp - 1))
468: continue;
1.51 kristaps 469: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 470: break;
471: }
1.17 kristaps 472:
1.42 kristaps 473: /* By default, assume a phrase. */
474: rc = ARGS_PHRASE;
475:
1.21 kristaps 476: /*
477: * Adjust new-buffer position to be beyond delimiter
478: * mark (e.g., Ta -> end + 2).
479: */
1.20 kristaps 480: if (p && pp) {
481: *pos += pp < p ? 2 : 1;
1.42 kristaps 482: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 483: p = pp < p ? pp : p;
484: } else if (p && ! pp) {
1.42 kristaps 485: rc = ARGS_PPHRASE;
1.20 kristaps 486: *pos += 1;
487: } else if (pp && ! p) {
1.17 kristaps 488: p = pp;
1.20 kristaps 489: *pos += 2;
1.44 kristaps 490: } else {
491: rc = ARGS_PEND;
1.20 kristaps 492: p = strchr(*v, 0);
1.44 kristaps 493: }
1.20 kristaps 494:
1.21 kristaps 495: /* Whitespace check for eoln case... */
1.76 kristaps 496: if ('\0' == *p && ' ' == *(p - 1))
1.67 kristaps 497: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.20 kristaps 498:
499: *pos += (int)(p - *v);
1.17 kristaps 500:
501: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 502: pp = p - 1;
503: while (pp > *v && ' ' == *pp) {
504: if (pp > *v && '\\' == *(pp - 1))
505: break;
506: pp--;
1.17 kristaps 507: }
1.20 kristaps 508: *(pp + 1) = 0;
1.17 kristaps 509:
1.20 kristaps 510: /* Strip delimiter's proceeding whitespace. */
511: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
512: /* Skip ahead. */ ;
1.1 kristaps 513:
1.41 kristaps 514: return(rc);
1.17 kristaps 515: }
1.1 kristaps 516:
1.17 kristaps 517: /*
518: * Process a quoted literal. A quote begins with a double-quote
519: * and ends with a double-quote NOT preceded by a double-quote.
520: * Whitespace is NOT involved in literal termination.
521: */
1.1 kristaps 522:
1.46 kristaps 523: if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
524: if ( ! (MDOC_PHRASELIT & m->flags))
525: *v = &buf[++(*pos)];
526:
1.47 kristaps 527: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 528: m->flags |= MDOC_PHRASELIT;
1.1 kristaps 529:
1.17 kristaps 530: for ( ; buf[*pos]; (*pos)++) {
531: if ('\"' != buf[*pos])
532: continue;
533: if ('\"' != buf[*pos + 1])
534: break;
535: (*pos)++;
536: }
1.1 kristaps 537:
1.46 kristaps 538: if ('\0' == buf[*pos]) {
1.76 kristaps 539: if (MDOC_PPHRASE & m->flags)
1.23 kristaps 540: return(ARGS_QWORD);
1.67 kristaps 541: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.18 kristaps 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:
1.76 kristaps 554: if ('\0' == buf[*pos])
1.67 kristaps 555: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 556:
1.17 kristaps 557: return(ARGS_QWORD);
1.1 kristaps 558: }
559:
1.75 kristaps 560: p = &buf[*pos];
1.76 kristaps 561: *v = mandoc_getarg(m->parse, &p, line, pos);
1.1 kristaps 562:
1.17 kristaps 563: return(ARGS_WORD);
1.64 kristaps 564: }
565:
566: /*
567: * Check if the string consists only of space-separated closing
1.65 kristaps 568: * delimiters. This is a bit of a dance: the first must be a close
569: * delimiter, but it may be followed by middle delimiters. Arbitrary
570: * whitespace may separate these tokens.
1.64 kristaps 571: */
572: static int
1.77 ! kristaps 573: args_checkpunct(const char *buf, int i)
1.64 kristaps 574: {
1.69 kristaps 575: int j;
576: char dbuf[DELIMSZ];
1.64 kristaps 577: enum mdelim d;
578:
1.65 kristaps 579: /* First token must be a close-delimiter. */
580:
1.69 kristaps 581: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
582: dbuf[j] = buf[i];
1.65 kristaps 583:
584: if (DELIMSZ == j)
585: return(0);
586:
1.69 kristaps 587: dbuf[j] = '\0';
1.70 kristaps 588: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.64 kristaps 589: return(0);
590:
1.69 kristaps 591: while (' ' == buf[i])
1.65 kristaps 592: i++;
593:
594: /* Remaining must NOT be open/none. */
595:
1.69 kristaps 596: while (buf[i]) {
1.65 kristaps 597: j = 0;
1.69 kristaps 598: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
599: dbuf[j++] = buf[i++];
1.65 kristaps 600:
601: if (DELIMSZ == j)
602: return(0);
603:
1.69 kristaps 604: dbuf[j] = '\0';
1.70 kristaps 605: d = mdoc_isdelim(dbuf);
1.64 kristaps 606: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 607: return(0);
608:
1.69 kristaps 609: while (' ' == buf[i])
1.64 kristaps 610: i++;
611: }
1.69 kristaps 612:
613: return('\0' == buf[i]);
1.1 kristaps 614: }
615:
1.63 kristaps 616: /*
617: * Match up an argument string (e.g., `-foo bar' having "foo") with the
618: * correrct identifier. It must apply to the given macro. If none was
619: * found (including bad matches), return MDOC_ARG_MAX.
620: */
1.57 kristaps 621: static enum mdocargt
1.35 kristaps 622: argv_a2arg(enum mdoct tok, const char *p)
1.1 kristaps 623: {
1.74 joerg 624: const enum mdocargt *argsp;
1.1 kristaps 625:
1.74 joerg 626: argsp = NULL;
1.1 kristaps 627:
628: switch (tok) {
629: case (MDOC_An):
1.74 joerg 630: argsp = args_An;
1.1 kristaps 631: break;
632: case (MDOC_Bd):
1.74 joerg 633: argsp = args_Bd;
1.1 kristaps 634: break;
635: case (MDOC_Bf):
1.74 joerg 636: argsp = args_Bf;
1.1 kristaps 637: break;
638: case (MDOC_Bk):
1.74 joerg 639: argsp = args_Bk;
1.1 kristaps 640: break;
641: case (MDOC_Bl):
1.74 joerg 642: argsp = args_Bl;
1.1 kristaps 643: break;
644: case (MDOC_Rv):
645: /* FALLTHROUGH */
646: case (MDOC_Ex):
1.74 joerg 647: argsp = args_Ex;
1.1 kristaps 648: break;
649: default:
1.73 kristaps 650: return(MDOC_ARG_MAX);
1.1 kristaps 651: }
1.63 kristaps 652:
1.74 joerg 653: assert(argsp);
1.73 kristaps 654:
1.74 joerg 655: for ( ; MDOC_ARG_MAX != *argsp ; argsp++)
656: if (0 == strcmp(p, mdoc_argnames[*argsp]))
657: return(*argsp);
1.1 kristaps 658:
659: return(MDOC_ARG_MAX);
660: }
661:
662: static int
1.11 kristaps 663: argv_multi(struct mdoc *m, int line,
1.1 kristaps 664: struct mdoc_argv *v, int *pos, char *buf)
665: {
1.43 kristaps 666: enum margserr ac;
1.1 kristaps 667: char *p;
668:
669: for (v->sz = 0; ; v->sz++) {
670: if ('-' == buf[*pos])
671: break;
1.77 ! kristaps 672: ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 673: if (ARGS_ERROR == ac)
1.1 kristaps 674: return(0);
1.43 kristaps 675: else if (ARGS_EOLN == ac)
1.1 kristaps 676: break;
677:
1.31 kristaps 678: if (0 == v->sz % MULTI_STEP)
679: v->value = mandoc_realloc(v->value,
1.1 kristaps 680: (v->sz + MULTI_STEP) * sizeof(char *));
1.31 kristaps 681:
682: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 683: }
684:
1.7 kristaps 685: return(1);
1.1 kristaps 686: }
687:
688: static int
1.11 kristaps 689: argv_opt_single(struct mdoc *m, int line,
1.1 kristaps 690: struct mdoc_argv *v, int *pos, char *buf)
691: {
1.43 kristaps 692: enum margserr ac;
1.1 kristaps 693: char *p;
694:
695: if ('-' == buf[*pos])
696: return(1);
697:
1.77 ! kristaps 698: ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 699: if (ARGS_ERROR == ac)
1.1 kristaps 700: return(0);
1.43 kristaps 701: if (ARGS_EOLN == ac)
1.1 kristaps 702: return(1);
703:
704: v->sz = 1;
1.31 kristaps 705: v->value = mandoc_malloc(sizeof(char *));
706: v->value[0] = mandoc_strdup(p);
1.2 kristaps 707:
1.1 kristaps 708: return(1);
709: }
710:
711: /*
712: * Parse a single, mandatory value from the stream.
713: */
714: static int
1.11 kristaps 715: argv_single(struct mdoc *m, int line,
1.1 kristaps 716: struct mdoc_argv *v, int *pos, char *buf)
717: {
1.43 kristaps 718: int ppos;
719: enum margserr ac;
1.1 kristaps 720: char *p;
721:
722: ppos = *pos;
723:
1.77 ! kristaps 724: ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
1.49 kristaps 725: if (ARGS_EOLN == ac) {
726: mdoc_pmsg(m, line, ppos, MANDOCERR_SYNTARGVCOUNT);
727: return(0);
728: } else if (ARGS_ERROR == ac)
1.1 kristaps 729: return(0);
730:
731: v->sz = 1;
1.31 kristaps 732: v->value = mandoc_malloc(sizeof(char *));
733: v->value[0] = mandoc_strdup(p);
1.2 kristaps 734:
1.1 kristaps 735: return(1);
736: }
737:
738: /*
739: * Determine rules for parsing arguments. Arguments can either accept
740: * no parameters, an optional single parameter, one parameter, or
741: * multiple parameters.
742: */
743: static int
744: argv(struct mdoc *mdoc, int line,
745: struct mdoc_argv *v, int *pos, char *buf)
746: {
747:
748: v->sz = 0;
749: v->value = NULL;
750:
1.68 kristaps 751: switch (argvflags[v->arg]) {
1.1 kristaps 752: case (ARGV_SINGLE):
753: return(argv_single(mdoc, line, v, pos, buf));
754: case (ARGV_MULTI):
755: return(argv_multi(mdoc, line, v, pos, buf));
756: case (ARGV_OPT_SINGLE):
757: return(argv_opt_single(mdoc, line, v, pos, buf));
1.68 kristaps 758: case (ARGV_NONE):
759: break;
1.1 kristaps 760: default:
1.68 kristaps 761: abort();
762: /* NOTREACHED */
1.1 kristaps 763: }
764:
765: return(1);
766: }
CVSweb