Annotation of mandoc/mdoc_argv.c, Revision 1.83
1.83 ! schwarze 1: /* $Id: mdoc_argv.c,v 1.82 2012/03/23 05:50:24 kristaps Exp $ */
1.1 kristaps 2: /*
1.81 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 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 <stdlib.h>
25: #include <stdio.h>
26: #include <string.h>
27:
1.72 kristaps 28: #include "mdoc.h"
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 */
1.76 kristaps 34: #define DELIMSZ 6 /* max possible size of a delimiter */
35:
36: enum argsflag {
37: ARGSFL_NONE = 0,
38: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
39: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
40: };
41:
42: enum argvflag {
43: ARGV_NONE, /* no args to flag (e.g., -split) */
44: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
1.83 ! schwarze 45: ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
1.76 kristaps 46: };
1.1 kristaps 47:
1.78 kristaps 48: struct mdocarg {
49: enum argsflag flags;
50: const enum mdocargt *argvs;
51: };
52:
1.80 kristaps 53: static void argn_free(struct mdoc_arg *, int);
1.39 kristaps 54: static enum margserr args(struct mdoc *, int, int *,
1.76 kristaps 55: char *, enum argsflag, char **);
1.77 kristaps 56: static int args_checkpunct(const char *, int);
1.80 kristaps 57: static int argv_multi(struct mdoc *, int,
1.1 kristaps 58: struct mdoc_argv *, int *, char *);
1.80 kristaps 59: static int argv_single(struct mdoc *, int,
1.1 kristaps 60: struct mdoc_argv *, int *, char *);
1.2 kristaps 61:
1.68 kristaps 62: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 63: ARGV_NONE, /* MDOC_Split */
64: ARGV_NONE, /* MDOC_Nosplit */
65: ARGV_NONE, /* MDOC_Ragged */
66: ARGV_NONE, /* MDOC_Unfilled */
67: ARGV_NONE, /* MDOC_Literal */
1.29 kristaps 68: ARGV_SINGLE, /* MDOC_File */
1.83 ! schwarze 69: ARGV_SINGLE, /* MDOC_Offset */
1.1 kristaps 70: ARGV_NONE, /* MDOC_Bullet */
71: ARGV_NONE, /* MDOC_Dash */
72: ARGV_NONE, /* MDOC_Hyphen */
73: ARGV_NONE, /* MDOC_Item */
74: ARGV_NONE, /* MDOC_Enum */
75: ARGV_NONE, /* MDOC_Tag */
76: ARGV_NONE, /* MDOC_Diag */
77: ARGV_NONE, /* MDOC_Hang */
78: ARGV_NONE, /* MDOC_Ohang */
79: ARGV_NONE, /* MDOC_Inset */
80: ARGV_MULTI, /* MDOC_Column */
1.83 ! schwarze 81: ARGV_SINGLE, /* MDOC_Width */
1.1 kristaps 82: ARGV_NONE, /* MDOC_Compact */
1.25 kristaps 83: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 84: ARGV_NONE, /* MDOC_Filled */
85: ARGV_NONE, /* MDOC_Words */
86: ARGV_NONE, /* MDOC_Emphasis */
87: ARGV_NONE, /* MDOC_Symbolic */
88: ARGV_NONE /* MDOC_Symbolic */
89: };
90:
1.73 kristaps 91: static const enum mdocargt args_Ex[] = {
92: MDOC_Std,
93: MDOC_ARG_MAX
94: };
95:
96: static const enum mdocargt args_An[] = {
97: MDOC_Split,
98: MDOC_Nosplit,
99: MDOC_ARG_MAX
100: };
101:
102: static const enum mdocargt args_Bd[] = {
103: MDOC_Ragged,
104: MDOC_Unfilled,
105: MDOC_Filled,
106: MDOC_Literal,
107: MDOC_File,
108: MDOC_Offset,
109: MDOC_Compact,
110: MDOC_Centred,
111: MDOC_ARG_MAX
112: };
113:
114: static const enum mdocargt args_Bf[] = {
115: MDOC_Emphasis,
116: MDOC_Literal,
117: MDOC_Symbolic,
118: MDOC_ARG_MAX
119: };
120:
121: static const enum mdocargt args_Bk[] = {
122: MDOC_Words,
123: MDOC_ARG_MAX
124: };
125:
126: static const enum mdocargt args_Bl[] = {
127: MDOC_Bullet,
128: MDOC_Dash,
129: MDOC_Hyphen,
130: MDOC_Item,
131: MDOC_Enum,
132: MDOC_Tag,
133: MDOC_Diag,
134: MDOC_Hang,
135: MDOC_Ohang,
136: MDOC_Inset,
137: MDOC_Column,
138: MDOC_Width,
139: MDOC_Offset,
140: MDOC_Compact,
141: MDOC_Nested,
142: MDOC_ARG_MAX
143: };
144:
1.78 kristaps 145: static const struct mdocarg mdocargs[MDOC_MAX] = {
146: { ARGSFL_NONE, NULL }, /* Ap */
147: { ARGSFL_NONE, NULL }, /* Dd */
148: { ARGSFL_NONE, NULL }, /* Dt */
149: { ARGSFL_NONE, NULL }, /* Os */
150: { ARGSFL_NONE, NULL }, /* Sh */
151: { ARGSFL_NONE, NULL }, /* Ss */
152: { ARGSFL_NONE, NULL }, /* Pp */
153: { ARGSFL_DELIM, NULL }, /* D1 */
154: { ARGSFL_DELIM, NULL }, /* Dl */
155: { ARGSFL_NONE, args_Bd }, /* Bd */
156: { ARGSFL_NONE, NULL }, /* Ed */
157: { ARGSFL_NONE, args_Bl }, /* Bl */
158: { ARGSFL_NONE, NULL }, /* El */
159: { ARGSFL_NONE, NULL }, /* It */
160: { ARGSFL_DELIM, NULL }, /* Ad */
161: { ARGSFL_DELIM, args_An }, /* An */
162: { ARGSFL_DELIM, NULL }, /* Ar */
163: { ARGSFL_NONE, NULL }, /* Cd */
164: { ARGSFL_DELIM, NULL }, /* Cm */
165: { ARGSFL_DELIM, NULL }, /* Dv */
166: { ARGSFL_DELIM, NULL }, /* Er */
167: { ARGSFL_DELIM, NULL }, /* Ev */
168: { ARGSFL_NONE, args_Ex }, /* Ex */
169: { ARGSFL_DELIM, NULL }, /* Fa */
170: { ARGSFL_NONE, NULL }, /* Fd */
171: { ARGSFL_DELIM, NULL }, /* Fl */
172: { ARGSFL_DELIM, NULL }, /* Fn */
173: { ARGSFL_DELIM, NULL }, /* Ft */
174: { ARGSFL_DELIM, NULL }, /* Ic */
175: { ARGSFL_NONE, NULL }, /* In */
176: { ARGSFL_DELIM, NULL }, /* Li */
177: { ARGSFL_NONE, NULL }, /* Nd */
178: { ARGSFL_DELIM, NULL }, /* Nm */
179: { ARGSFL_DELIM, NULL }, /* Op */
180: { ARGSFL_NONE, NULL }, /* Ot */
181: { ARGSFL_DELIM, NULL }, /* Pa */
182: { ARGSFL_NONE, args_Ex }, /* Rv */
183: { ARGSFL_DELIM, NULL }, /* St */
184: { ARGSFL_DELIM, NULL }, /* Va */
185: { ARGSFL_DELIM, NULL }, /* Vt */
186: { ARGSFL_DELIM, NULL }, /* Xr */
187: { ARGSFL_NONE, NULL }, /* %A */
188: { ARGSFL_NONE, NULL }, /* %B */
189: { ARGSFL_NONE, NULL }, /* %D */
190: { ARGSFL_NONE, NULL }, /* %I */
191: { ARGSFL_NONE, NULL }, /* %J */
192: { ARGSFL_NONE, NULL }, /* %N */
193: { ARGSFL_NONE, NULL }, /* %O */
194: { ARGSFL_NONE, NULL }, /* %P */
195: { ARGSFL_NONE, NULL }, /* %R */
196: { ARGSFL_NONE, NULL }, /* %T */
197: { ARGSFL_NONE, NULL }, /* %V */
198: { ARGSFL_DELIM, NULL }, /* Ac */
199: { ARGSFL_NONE, NULL }, /* Ao */
200: { ARGSFL_DELIM, NULL }, /* Aq */
201: { ARGSFL_DELIM, NULL }, /* At */
202: { ARGSFL_DELIM, NULL }, /* Bc */
203: { ARGSFL_NONE, args_Bf }, /* Bf */
204: { ARGSFL_NONE, NULL }, /* Bo */
205: { ARGSFL_DELIM, NULL }, /* Bq */
206: { ARGSFL_DELIM, NULL }, /* Bsx */
207: { ARGSFL_DELIM, NULL }, /* Bx */
208: { ARGSFL_NONE, NULL }, /* Db */
209: { ARGSFL_DELIM, NULL }, /* Dc */
210: { ARGSFL_NONE, NULL }, /* Do */
211: { ARGSFL_DELIM, NULL }, /* Dq */
212: { ARGSFL_DELIM, NULL }, /* Ec */
213: { ARGSFL_NONE, NULL }, /* Ef */
214: { ARGSFL_DELIM, NULL }, /* Em */
215: { ARGSFL_NONE, NULL }, /* Eo */
216: { ARGSFL_DELIM, NULL }, /* Fx */
217: { ARGSFL_DELIM, NULL }, /* Ms */
218: { ARGSFL_DELIM, NULL }, /* No */
219: { ARGSFL_DELIM, NULL }, /* Ns */
220: { ARGSFL_DELIM, NULL }, /* Nx */
221: { ARGSFL_DELIM, NULL }, /* Ox */
222: { ARGSFL_DELIM, NULL }, /* Pc */
223: { ARGSFL_DELIM, NULL }, /* Pf */
224: { ARGSFL_NONE, NULL }, /* Po */
225: { ARGSFL_DELIM, NULL }, /* Pq */
226: { ARGSFL_DELIM, NULL }, /* Qc */
227: { ARGSFL_DELIM, NULL }, /* Ql */
228: { ARGSFL_NONE, NULL }, /* Qo */
229: { ARGSFL_DELIM, NULL }, /* Qq */
230: { ARGSFL_NONE, NULL }, /* Re */
231: { ARGSFL_NONE, NULL }, /* Rs */
232: { ARGSFL_DELIM, NULL }, /* Sc */
233: { ARGSFL_NONE, NULL }, /* So */
234: { ARGSFL_DELIM, NULL }, /* Sq */
235: { ARGSFL_NONE, NULL }, /* Sm */
236: { ARGSFL_DELIM, NULL }, /* Sx */
237: { ARGSFL_DELIM, NULL }, /* Sy */
238: { ARGSFL_DELIM, NULL }, /* Tn */
239: { ARGSFL_DELIM, NULL }, /* Ux */
240: { ARGSFL_DELIM, NULL }, /* Xc */
241: { ARGSFL_NONE, NULL }, /* Xo */
242: { ARGSFL_NONE, NULL }, /* Fo */
243: { ARGSFL_NONE, NULL }, /* Fc */
244: { ARGSFL_NONE, NULL }, /* Oo */
245: { ARGSFL_DELIM, NULL }, /* Oc */
246: { ARGSFL_NONE, args_Bk }, /* Bk */
247: { ARGSFL_NONE, NULL }, /* Ek */
248: { ARGSFL_NONE, NULL }, /* Bt */
249: { ARGSFL_NONE, NULL }, /* Hf */
250: { ARGSFL_NONE, NULL }, /* Fr */
251: { ARGSFL_NONE, NULL }, /* Ud */
252: { ARGSFL_NONE, NULL }, /* Lb */
253: { ARGSFL_NONE, NULL }, /* Lp */
254: { ARGSFL_DELIM, NULL }, /* Lk */
255: { ARGSFL_DELIM, NULL }, /* Mt */
256: { ARGSFL_DELIM, NULL }, /* Brq */
257: { ARGSFL_NONE, NULL }, /* Bro */
258: { ARGSFL_DELIM, NULL }, /* Brc */
259: { ARGSFL_NONE, NULL }, /* %C */
260: { ARGSFL_NONE, NULL }, /* Es */
261: { ARGSFL_NONE, NULL }, /* En */
262: { ARGSFL_NONE, NULL }, /* Dx */
263: { ARGSFL_NONE, NULL }, /* %Q */
264: { ARGSFL_NONE, NULL }, /* br */
265: { ARGSFL_NONE, NULL }, /* sp */
266: { ARGSFL_NONE, NULL }, /* %U */
267: { ARGSFL_NONE, NULL }, /* Ta */
268: };
269:
270:
1.1 kristaps 271: /*
272: * Parse an argument from line text. This comes in the form of -key
273: * [value0...], which may either have a single mandatory value, at least
274: * one mandatory value, an optional single value, or no value.
275: */
1.40 kristaps 276: enum margverr
1.35 kristaps 277: mdoc_argv(struct mdoc *m, int line, enum mdoct tok,
1.1 kristaps 278: struct mdoc_arg **v, int *pos, char *buf)
279: {
280: char *p, sv;
281: struct mdoc_argv tmp;
282: struct mdoc_arg *arg;
1.78 kristaps 283: const enum mdocargt *ap;
1.1 kristaps 284:
1.52 kristaps 285: if ('\0' == buf[*pos])
1.1 kristaps 286: return(ARGV_EOLN);
1.78 kristaps 287: else if (NULL == (ap = mdocargs[tok].argvs))
288: return(ARGV_WORD);
1.80 kristaps 289: else if ('-' != buf[*pos])
290: return(ARGV_WORD);
1.1 kristaps 291:
1.78 kristaps 292: /* Seek to the first unescaped space. */
1.1 kristaps 293:
294: p = &buf[++(*pos)];
295:
296: assert(*pos > 0);
297:
1.78 kristaps 298: for ( ; buf[*pos] ; (*pos)++)
299: if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
300: break;
1.1 kristaps 301:
1.78 kristaps 302: /*
303: * We want to nil-terminate the word to look it up (it's easier
304: * that way). But we may not have a flag, in which case we need
305: * to restore the line as-is. So keep around the stray byte,
306: * which we'll reset upon exiting (if necessary).
307: */
1.1 kristaps 308:
1.78 kristaps 309: if ('\0' != (sv = buf[*pos]))
1.52 kristaps 310: buf[(*pos)++] = '\0';
1.78 kristaps 311:
312: /*
313: * Now look up the word as a flag. Use temporary storage that
314: * we'll copy into the node's flags, if necessary.
315: */
1.1 kristaps 316:
1.68 kristaps 317: memset(&tmp, 0, sizeof(struct mdoc_argv));
1.78 kristaps 318:
1.1 kristaps 319: tmp.line = line;
320: tmp.pos = *pos;
1.78 kristaps 321: tmp.arg = MDOC_ARG_MAX;
1.1 kristaps 322:
1.78 kristaps 323: while (MDOC_ARG_MAX != (tmp.arg = *ap++))
324: if (0 == strcmp(p, mdoc_argnames[tmp.arg]))
325: break;
1.1 kristaps 326:
1.78 kristaps 327: if (MDOC_ARG_MAX == tmp.arg) {
328: /*
329: * The flag was not found.
330: * Restore saved zeroed byte and return as a word.
331: */
1.1 kristaps 332: if (sv)
333: buf[*pos - 1] = sv;
334: return(ARGV_WORD);
335: }
336:
1.78 kristaps 337: /* Read to the next word (the argument). */
338:
1.1 kristaps 339: while (buf[*pos] && ' ' == buf[*pos])
340: (*pos)++;
341:
1.79 kristaps 342: switch (argvflags[tmp.arg]) {
343: case (ARGV_SINGLE):
344: if ( ! argv_single(m, line, &tmp, pos, buf))
345: return(ARGV_ERROR);
346: break;
347: case (ARGV_MULTI):
348: if ( ! argv_multi(m, line, &tmp, pos, buf))
349: return(ARGV_ERROR);
350: break;
351: case (ARGV_NONE):
352: break;
353: }
1.1 kristaps 354:
1.31 kristaps 355: if (NULL == (arg = *v))
1.32 kristaps 356: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 357:
358: arg->argc++;
1.31 kristaps 359: arg->argv = mandoc_realloc
360: (arg->argv, arg->argc * sizeof(struct mdoc_argv));
1.2 kristaps 361:
1.68 kristaps 362: memcpy(&arg->argv[(int)arg->argc - 1],
1.1 kristaps 363: &tmp, sizeof(struct mdoc_argv));
364:
365: return(ARGV_ARG);
366: }
367:
368: void
369: mdoc_argv_free(struct mdoc_arg *p)
370: {
1.37 kristaps 371: int i;
1.1 kristaps 372:
373: if (NULL == p)
374: return;
375:
376: if (p->refcnt) {
377: --(p->refcnt);
378: if (p->refcnt)
379: return;
380: }
381: assert(p->argc);
382:
1.37 kristaps 383: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 384: argn_free(p, i);
1.37 kristaps 385:
386: free(p->argv);
387: free(p);
388: }
389:
1.71 kristaps 390: static void
391: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 392: {
1.58 kristaps 393: struct mdoc_argv *arg;
1.37 kristaps 394: int j;
1.58 kristaps 395:
396: arg = &p->argv[iarg];
1.1 kristaps 397:
1.37 kristaps 398: if (arg->sz && arg->value) {
399: for (j = (int)arg->sz - 1; j >= 0; j--)
400: free(arg->value[j]);
401: free(arg->value);
1.1 kristaps 402: }
403:
1.37 kristaps 404: for (--p->argc; iarg < (int)p->argc; iarg++)
405: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 406: }
407:
1.39 kristaps 408: enum margserr
1.76 kristaps 409: mdoc_zargs(struct mdoc *m, int line, int *pos, char *buf, char **v)
1.19 kristaps 410: {
411:
1.76 kristaps 412: return(args(m, line, pos, buf, ARGSFL_NONE, v));
1.19 kristaps 413: }
414:
1.39 kristaps 415: enum margserr
1.35 kristaps 416: mdoc_args(struct mdoc *m, int line, int *pos,
417: char *buf, enum mdoct tok, char **v)
1.1 kristaps 418: {
1.76 kristaps 419: enum argsflag fl;
1.1 kristaps 420: struct mdoc_node *n;
421:
1.78 kristaps 422: fl = mdocargs[tok].flags;
1.1 kristaps 423:
1.17 kristaps 424: if (MDOC_It != tok)
425: return(args(m, line, pos, buf, fl, v));
426:
1.51 kristaps 427: /*
428: * We know that we're in an `It', so it's reasonable to expect
429: * us to be sitting in a `Bl'. Someday this may not be the case
430: * (if we allow random `It's sitting out there), so provide a
431: * safe fall-back into the default behaviour.
432: */
433:
1.17 kristaps 434: for (n = m->last; n; n = n->parent)
1.50 kristaps 435: if (MDOC_Bl == n->tok)
1.76 kristaps 436: if (LIST_column == n->norm->Bl.type) {
437: fl = ARGSFL_TABSEP;
438: break;
439: }
1.1 kristaps 440:
1.11 kristaps 441: return(args(m, line, pos, buf, fl, v));
1.1 kristaps 442: }
443:
1.39 kristaps 444: static enum margserr
1.13 kristaps 445: args(struct mdoc *m, int line, int *pos,
1.76 kristaps 446: char *buf, enum argsflag fl, char **v)
1.1 kristaps 447: {
1.52 kristaps 448: char *p, *pp;
449: enum margserr rc;
1.1 kristaps 450:
1.46 kristaps 451: if ('\0' == buf[*pos]) {
1.47 kristaps 452: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 453: return(ARGS_EOLN);
454: /*
455: * If we're not in a partial phrase and the flag for
456: * being a phrase literal is still set, the punctuation
457: * is unterminated.
458: */
459: if (MDOC_PHRASELIT & m->flags)
1.67 kristaps 460: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.46 kristaps 461:
462: m->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 463: return(ARGS_EOLN);
1.46 kristaps 464: }
1.1 kristaps 465:
1.64 kristaps 466: *v = &buf[*pos];
1.17 kristaps 467:
1.76 kristaps 468: if (ARGSFL_DELIM == fl)
1.77 kristaps 469: if (args_checkpunct(buf, *pos))
1.76 kristaps 470: return(ARGS_PUNCT);
1.1 kristaps 471:
1.17 kristaps 472: /*
473: * First handle TABSEP items, restricted to `Bl -column'. This
474: * ignores conventional token parsing and instead uses tabs or
475: * `Ta' macros to separate phrases. Phrases are parsed again
476: * for arguments at a later phase.
477: */
1.1 kristaps 478:
1.76 kristaps 479: if (ARGSFL_TABSEP == fl) {
1.20 kristaps 480: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 481: p = strchr(*v, '\t');
1.46 kristaps 482: pp = NULL;
1.17 kristaps 483:
484: /* Scan ahead to unescaped `Ta'. */
1.46 kristaps 485: if ( ! (MDOC_PHRASELIT & m->flags))
486: for (pp = *v; ; pp++) {
487: if (NULL == (pp = strstr(pp, "Ta")))
488: break;
489: if (pp > *v && ' ' != *(pp - 1))
490: continue;
1.51 kristaps 491: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 492: break;
493: }
1.17 kristaps 494:
1.42 kristaps 495: /* By default, assume a phrase. */
496: rc = ARGS_PHRASE;
497:
1.21 kristaps 498: /*
499: * Adjust new-buffer position to be beyond delimiter
500: * mark (e.g., Ta -> end + 2).
501: */
1.20 kristaps 502: if (p && pp) {
503: *pos += pp < p ? 2 : 1;
1.42 kristaps 504: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 505: p = pp < p ? pp : p;
506: } else if (p && ! pp) {
1.42 kristaps 507: rc = ARGS_PPHRASE;
1.20 kristaps 508: *pos += 1;
509: } else if (pp && ! p) {
1.17 kristaps 510: p = pp;
1.20 kristaps 511: *pos += 2;
1.44 kristaps 512: } else {
513: rc = ARGS_PEND;
1.20 kristaps 514: p = strchr(*v, 0);
1.44 kristaps 515: }
1.20 kristaps 516:
1.21 kristaps 517: /* Whitespace check for eoln case... */
1.76 kristaps 518: if ('\0' == *p && ' ' == *(p - 1))
1.67 kristaps 519: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.20 kristaps 520:
521: *pos += (int)(p - *v);
1.17 kristaps 522:
523: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 524: pp = p - 1;
525: while (pp > *v && ' ' == *pp) {
526: if (pp > *v && '\\' == *(pp - 1))
527: break;
528: pp--;
1.17 kristaps 529: }
1.20 kristaps 530: *(pp + 1) = 0;
1.17 kristaps 531:
1.20 kristaps 532: /* Strip delimiter's proceeding whitespace. */
533: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
534: /* Skip ahead. */ ;
1.1 kristaps 535:
1.41 kristaps 536: return(rc);
1.17 kristaps 537: }
1.1 kristaps 538:
1.17 kristaps 539: /*
540: * Process a quoted literal. A quote begins with a double-quote
541: * and ends with a double-quote NOT preceded by a double-quote.
542: * Whitespace is NOT involved in literal termination.
543: */
1.1 kristaps 544:
1.46 kristaps 545: if (MDOC_PHRASELIT & m->flags || '\"' == buf[*pos]) {
546: if ( ! (MDOC_PHRASELIT & m->flags))
547: *v = &buf[++(*pos)];
548:
1.47 kristaps 549: if (MDOC_PPHRASE & m->flags)
1.46 kristaps 550: m->flags |= MDOC_PHRASELIT;
1.1 kristaps 551:
1.17 kristaps 552: for ( ; buf[*pos]; (*pos)++) {
553: if ('\"' != buf[*pos])
554: continue;
555: if ('\"' != buf[*pos + 1])
556: break;
557: (*pos)++;
558: }
1.1 kristaps 559:
1.46 kristaps 560: if ('\0' == buf[*pos]) {
1.76 kristaps 561: if (MDOC_PPHRASE & m->flags)
1.23 kristaps 562: return(ARGS_QWORD);
1.67 kristaps 563: mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE);
1.18 kristaps 564: return(ARGS_QWORD);
1.17 kristaps 565: }
1.1 kristaps 566:
1.46 kristaps 567: m->flags &= ~MDOC_PHRASELIT;
568: buf[(*pos)++] = '\0';
1.1 kristaps 569:
1.46 kristaps 570: if ('\0' == buf[*pos])
1.17 kristaps 571: return(ARGS_QWORD);
1.1 kristaps 572:
1.17 kristaps 573: while (' ' == buf[*pos])
574: (*pos)++;
1.1 kristaps 575:
1.76 kristaps 576: if ('\0' == buf[*pos])
1.67 kristaps 577: mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 578:
1.17 kristaps 579: return(ARGS_QWORD);
1.1 kristaps 580: }
581:
1.75 kristaps 582: p = &buf[*pos];
1.76 kristaps 583: *v = mandoc_getarg(m->parse, &p, line, pos);
1.1 kristaps 584:
1.17 kristaps 585: return(ARGS_WORD);
1.64 kristaps 586: }
587:
588: /*
589: * Check if the string consists only of space-separated closing
1.65 kristaps 590: * delimiters. This is a bit of a dance: the first must be a close
591: * delimiter, but it may be followed by middle delimiters. Arbitrary
592: * whitespace may separate these tokens.
1.64 kristaps 593: */
594: static int
1.77 kristaps 595: args_checkpunct(const char *buf, int i)
1.64 kristaps 596: {
1.69 kristaps 597: int j;
598: char dbuf[DELIMSZ];
1.64 kristaps 599: enum mdelim d;
600:
1.65 kristaps 601: /* First token must be a close-delimiter. */
602:
1.69 kristaps 603: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
604: dbuf[j] = buf[i];
1.65 kristaps 605:
606: if (DELIMSZ == j)
607: return(0);
608:
1.69 kristaps 609: dbuf[j] = '\0';
1.70 kristaps 610: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.64 kristaps 611: return(0);
612:
1.69 kristaps 613: while (' ' == buf[i])
1.65 kristaps 614: i++;
615:
616: /* Remaining must NOT be open/none. */
617:
1.69 kristaps 618: while (buf[i]) {
1.65 kristaps 619: j = 0;
1.69 kristaps 620: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
621: dbuf[j++] = buf[i++];
1.65 kristaps 622:
623: if (DELIMSZ == j)
624: return(0);
625:
1.69 kristaps 626: dbuf[j] = '\0';
1.70 kristaps 627: d = mdoc_isdelim(dbuf);
1.64 kristaps 628: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 629: return(0);
630:
1.69 kristaps 631: while (' ' == buf[i])
1.64 kristaps 632: i++;
633: }
1.69 kristaps 634:
635: return('\0' == buf[i]);
1.1 kristaps 636: }
637:
638: static int
1.11 kristaps 639: argv_multi(struct mdoc *m, int line,
1.1 kristaps 640: struct mdoc_argv *v, int *pos, char *buf)
641: {
1.43 kristaps 642: enum margserr ac;
1.1 kristaps 643: char *p;
644:
645: for (v->sz = 0; ; v->sz++) {
646: if ('-' == buf[*pos])
647: break;
1.77 kristaps 648: ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 649: if (ARGS_ERROR == ac)
1.1 kristaps 650: return(0);
1.43 kristaps 651: else if (ARGS_EOLN == ac)
1.1 kristaps 652: break;
653:
1.31 kristaps 654: if (0 == v->sz % MULTI_STEP)
655: v->value = mandoc_realloc(v->value,
1.1 kristaps 656: (v->sz + MULTI_STEP) * sizeof(char *));
1.31 kristaps 657:
658: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 659: }
660:
1.7 kristaps 661: return(1);
1.1 kristaps 662: }
663:
664: static int
1.83 ! schwarze 665: argv_single(struct mdoc *m, int line,
1.1 kristaps 666: struct mdoc_argv *v, int *pos, char *buf)
667: {
1.43 kristaps 668: enum margserr ac;
1.1 kristaps 669: char *p;
670:
1.77 kristaps 671: ac = args(m, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 672: if (ARGS_ERROR == ac)
1.1 kristaps 673: return(0);
1.43 kristaps 674: if (ARGS_EOLN == ac)
1.1 kristaps 675: return(1);
676:
677: v->sz = 1;
1.31 kristaps 678: v->value = mandoc_malloc(sizeof(char *));
679: v->value[0] = mandoc_strdup(p);
1.1 kristaps 680:
681: return(1);
682: }
CVSweb