Annotation of mandoc/mdoc_argv.c, Revision 1.87
1.87 ! schwarze 1: /* $Id: mdoc_argv.c,v 1.86 2012/11/18 00:05:35 schwarze Exp $ */
1.1 kristaps 2: /*
1.81 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.84 schwarze 4: * Copyright (c) 2012 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.3 kristaps 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.3 kristaps 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 17: */
1.33 kristaps 18: #ifdef HAVE_CONFIG_H
19: #include "config.h"
20: #endif
21:
1.1 kristaps 22: #include <sys/types.h>
23:
24: #include <assert.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) */
1.83 schwarze 46: ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
1.76 kristaps 47: };
1.1 kristaps 48:
1.78 kristaps 49: struct mdocarg {
50: enum argsflag flags;
51: const enum mdocargt *argvs;
52: };
53:
1.80 kristaps 54: static void argn_free(struct mdoc_arg *, int);
1.39 kristaps 55: static enum margserr args(struct mdoc *, int, int *,
1.76 kristaps 56: char *, enum argsflag, char **);
1.77 kristaps 57: static int args_checkpunct(const char *, int);
1.80 kristaps 58: static int argv_multi(struct mdoc *, int,
1.1 kristaps 59: struct mdoc_argv *, int *, char *);
1.80 kristaps 60: static int argv_single(struct mdoc *, int,
1.1 kristaps 61: struct mdoc_argv *, int *, char *);
1.2 kristaps 62:
1.68 kristaps 63: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 64: ARGV_NONE, /* MDOC_Split */
65: ARGV_NONE, /* MDOC_Nosplit */
66: ARGV_NONE, /* MDOC_Ragged */
67: ARGV_NONE, /* MDOC_Unfilled */
68: ARGV_NONE, /* MDOC_Literal */
1.29 kristaps 69: ARGV_SINGLE, /* MDOC_File */
1.83 schwarze 70: ARGV_SINGLE, /* MDOC_Offset */
1.1 kristaps 71: ARGV_NONE, /* MDOC_Bullet */
72: ARGV_NONE, /* MDOC_Dash */
73: ARGV_NONE, /* MDOC_Hyphen */
74: ARGV_NONE, /* MDOC_Item */
75: ARGV_NONE, /* MDOC_Enum */
76: ARGV_NONE, /* MDOC_Tag */
77: ARGV_NONE, /* MDOC_Diag */
78: ARGV_NONE, /* MDOC_Hang */
79: ARGV_NONE, /* MDOC_Ohang */
80: ARGV_NONE, /* MDOC_Inset */
81: ARGV_MULTI, /* MDOC_Column */
1.83 schwarze 82: ARGV_SINGLE, /* MDOC_Width */
1.1 kristaps 83: ARGV_NONE, /* MDOC_Compact */
1.25 kristaps 84: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 85: ARGV_NONE, /* MDOC_Filled */
86: ARGV_NONE, /* MDOC_Words */
87: ARGV_NONE, /* MDOC_Emphasis */
88: ARGV_NONE, /* MDOC_Symbolic */
89: ARGV_NONE /* MDOC_Symbolic */
90: };
91:
1.73 kristaps 92: static const enum mdocargt args_Ex[] = {
93: MDOC_Std,
94: MDOC_ARG_MAX
95: };
96:
97: static const enum mdocargt args_An[] = {
98: MDOC_Split,
99: MDOC_Nosplit,
100: MDOC_ARG_MAX
101: };
102:
103: static const enum mdocargt args_Bd[] = {
104: MDOC_Ragged,
105: MDOC_Unfilled,
106: MDOC_Filled,
107: MDOC_Literal,
108: MDOC_File,
109: MDOC_Offset,
110: MDOC_Compact,
111: MDOC_Centred,
112: MDOC_ARG_MAX
113: };
114:
115: static const enum mdocargt args_Bf[] = {
116: MDOC_Emphasis,
117: MDOC_Literal,
118: MDOC_Symbolic,
119: MDOC_ARG_MAX
120: };
121:
122: static const enum mdocargt args_Bk[] = {
123: MDOC_Words,
124: MDOC_ARG_MAX
125: };
126:
127: static const enum mdocargt args_Bl[] = {
128: MDOC_Bullet,
129: MDOC_Dash,
130: MDOC_Hyphen,
131: MDOC_Item,
132: MDOC_Enum,
133: MDOC_Tag,
134: MDOC_Diag,
135: MDOC_Hang,
136: MDOC_Ohang,
137: MDOC_Inset,
138: MDOC_Column,
139: MDOC_Width,
140: MDOC_Offset,
141: MDOC_Compact,
142: MDOC_Nested,
143: MDOC_ARG_MAX
144: };
145:
1.78 kristaps 146: static const struct mdocarg mdocargs[MDOC_MAX] = {
1.84 schwarze 147: { ARGSFL_DELIM, NULL }, /* Ap */
1.78 kristaps 148: { ARGSFL_NONE, NULL }, /* Dd */
149: { ARGSFL_NONE, NULL }, /* Dt */
150: { ARGSFL_NONE, NULL }, /* Os */
151: { ARGSFL_NONE, NULL }, /* Sh */
152: { ARGSFL_NONE, NULL }, /* Ss */
153: { ARGSFL_NONE, NULL }, /* Pp */
154: { ARGSFL_DELIM, NULL }, /* D1 */
155: { ARGSFL_DELIM, NULL }, /* Dl */
156: { ARGSFL_NONE, args_Bd }, /* Bd */
157: { ARGSFL_NONE, NULL }, /* Ed */
158: { ARGSFL_NONE, args_Bl }, /* Bl */
159: { ARGSFL_NONE, NULL }, /* El */
160: { ARGSFL_NONE, NULL }, /* It */
161: { ARGSFL_DELIM, NULL }, /* Ad */
162: { ARGSFL_DELIM, args_An }, /* An */
163: { ARGSFL_DELIM, NULL }, /* Ar */
1.86 schwarze 164: { ARGSFL_DELIM, NULL }, /* Cd */
1.78 kristaps 165: { ARGSFL_DELIM, NULL }, /* Cm */
166: { ARGSFL_DELIM, NULL }, /* Dv */
167: { ARGSFL_DELIM, NULL }, /* Er */
168: { ARGSFL_DELIM, NULL }, /* Ev */
169: { ARGSFL_NONE, args_Ex }, /* Ex */
170: { ARGSFL_DELIM, NULL }, /* Fa */
171: { ARGSFL_NONE, NULL }, /* Fd */
172: { ARGSFL_DELIM, NULL }, /* Fl */
173: { ARGSFL_DELIM, NULL }, /* Fn */
174: { ARGSFL_DELIM, NULL }, /* Ft */
175: { ARGSFL_DELIM, NULL }, /* Ic */
1.84 schwarze 176: { ARGSFL_DELIM, NULL }, /* In */
1.78 kristaps 177: { ARGSFL_DELIM, NULL }, /* Li */
178: { ARGSFL_NONE, NULL }, /* Nd */
179: { ARGSFL_DELIM, NULL }, /* Nm */
180: { ARGSFL_DELIM, NULL }, /* Op */
181: { ARGSFL_NONE, NULL }, /* Ot */
182: { ARGSFL_DELIM, NULL }, /* Pa */
183: { ARGSFL_NONE, args_Ex }, /* Rv */
184: { ARGSFL_DELIM, NULL }, /* St */
185: { ARGSFL_DELIM, NULL }, /* Va */
186: { ARGSFL_DELIM, NULL }, /* Vt */
187: { ARGSFL_DELIM, NULL }, /* Xr */
188: { ARGSFL_NONE, NULL }, /* %A */
189: { ARGSFL_NONE, NULL }, /* %B */
190: { ARGSFL_NONE, NULL }, /* %D */
191: { ARGSFL_NONE, NULL }, /* %I */
192: { ARGSFL_NONE, NULL }, /* %J */
193: { ARGSFL_NONE, NULL }, /* %N */
194: { ARGSFL_NONE, NULL }, /* %O */
195: { ARGSFL_NONE, NULL }, /* %P */
196: { ARGSFL_NONE, NULL }, /* %R */
197: { ARGSFL_NONE, NULL }, /* %T */
198: { ARGSFL_NONE, NULL }, /* %V */
199: { ARGSFL_DELIM, NULL }, /* Ac */
200: { ARGSFL_NONE, NULL }, /* Ao */
201: { ARGSFL_DELIM, NULL }, /* Aq */
202: { ARGSFL_DELIM, NULL }, /* At */
203: { ARGSFL_DELIM, NULL }, /* Bc */
204: { ARGSFL_NONE, args_Bf }, /* Bf */
205: { ARGSFL_NONE, NULL }, /* Bo */
206: { ARGSFL_DELIM, NULL }, /* Bq */
207: { ARGSFL_DELIM, NULL }, /* Bsx */
208: { ARGSFL_DELIM, NULL }, /* Bx */
209: { ARGSFL_NONE, NULL }, /* Db */
210: { ARGSFL_DELIM, NULL }, /* Dc */
211: { ARGSFL_NONE, NULL }, /* Do */
212: { ARGSFL_DELIM, NULL }, /* Dq */
213: { ARGSFL_DELIM, NULL }, /* Ec */
214: { ARGSFL_NONE, NULL }, /* Ef */
215: { ARGSFL_DELIM, NULL }, /* Em */
216: { ARGSFL_NONE, NULL }, /* Eo */
217: { ARGSFL_DELIM, NULL }, /* Fx */
218: { ARGSFL_DELIM, NULL }, /* Ms */
219: { ARGSFL_DELIM, NULL }, /* No */
220: { ARGSFL_DELIM, NULL }, /* Ns */
221: { ARGSFL_DELIM, NULL }, /* Nx */
222: { ARGSFL_DELIM, NULL }, /* Ox */
223: { ARGSFL_DELIM, NULL }, /* Pc */
224: { ARGSFL_DELIM, NULL }, /* Pf */
225: { ARGSFL_NONE, NULL }, /* Po */
226: { ARGSFL_DELIM, NULL }, /* Pq */
227: { ARGSFL_DELIM, NULL }, /* Qc */
228: { ARGSFL_DELIM, NULL }, /* Ql */
229: { ARGSFL_NONE, NULL }, /* Qo */
230: { ARGSFL_DELIM, NULL }, /* Qq */
231: { ARGSFL_NONE, NULL }, /* Re */
232: { ARGSFL_NONE, NULL }, /* Rs */
233: { ARGSFL_DELIM, NULL }, /* Sc */
234: { ARGSFL_NONE, NULL }, /* So */
235: { ARGSFL_DELIM, NULL }, /* Sq */
236: { ARGSFL_NONE, NULL }, /* Sm */
237: { ARGSFL_DELIM, NULL }, /* Sx */
238: { ARGSFL_DELIM, NULL }, /* Sy */
239: { ARGSFL_DELIM, NULL }, /* Tn */
240: { ARGSFL_DELIM, NULL }, /* Ux */
241: { ARGSFL_DELIM, NULL }, /* Xc */
242: { ARGSFL_NONE, NULL }, /* Xo */
243: { ARGSFL_NONE, NULL }, /* Fo */
1.86 schwarze 244: { ARGSFL_DELIM, NULL }, /* Fc */
1.78 kristaps 245: { ARGSFL_NONE, NULL }, /* Oo */
246: { ARGSFL_DELIM, NULL }, /* Oc */
247: { ARGSFL_NONE, args_Bk }, /* Bk */
248: { ARGSFL_NONE, NULL }, /* Ek */
249: { ARGSFL_NONE, NULL }, /* Bt */
250: { ARGSFL_NONE, NULL }, /* Hf */
251: { ARGSFL_NONE, NULL }, /* Fr */
252: { ARGSFL_NONE, NULL }, /* Ud */
1.86 schwarze 253: { ARGSFL_DELIM, NULL }, /* Lb */
1.78 kristaps 254: { ARGSFL_NONE, NULL }, /* Lp */
255: { ARGSFL_DELIM, NULL }, /* Lk */
256: { ARGSFL_DELIM, NULL }, /* Mt */
257: { ARGSFL_DELIM, NULL }, /* Brq */
258: { ARGSFL_NONE, NULL }, /* Bro */
259: { ARGSFL_DELIM, NULL }, /* Brc */
260: { ARGSFL_NONE, NULL }, /* %C */
261: { ARGSFL_NONE, NULL }, /* Es */
262: { ARGSFL_NONE, NULL }, /* En */
263: { ARGSFL_NONE, NULL }, /* Dx */
264: { ARGSFL_NONE, NULL }, /* %Q */
265: { ARGSFL_NONE, NULL }, /* br */
266: { ARGSFL_NONE, NULL }, /* sp */
267: { ARGSFL_NONE, NULL }, /* %U */
268: { ARGSFL_NONE, NULL }, /* Ta */
269: };
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
1.85 schwarze 278: mdoc_argv(struct mdoc *mdoc, 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;
1.78 kristaps 284: const enum mdocargt *ap;
1.1 kristaps 285:
1.52 kristaps 286: if ('\0' == buf[*pos])
1.1 kristaps 287: return(ARGV_EOLN);
1.78 kristaps 288: else if (NULL == (ap = mdocargs[tok].argvs))
289: return(ARGV_WORD);
1.80 kristaps 290: else if ('-' != buf[*pos])
291: return(ARGV_WORD);
1.1 kristaps 292:
1.78 kristaps 293: /* Seek to the first unescaped space. */
1.1 kristaps 294:
295: p = &buf[++(*pos)];
296:
297: assert(*pos > 0);
298:
1.78 kristaps 299: for ( ; buf[*pos] ; (*pos)++)
300: if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
301: break;
1.1 kristaps 302:
1.78 kristaps 303: /*
304: * We want to nil-terminate the word to look it up (it's easier
305: * that way). But we may not have a flag, in which case we need
306: * to restore the line as-is. So keep around the stray byte,
307: * which we'll reset upon exiting (if necessary).
308: */
1.1 kristaps 309:
1.78 kristaps 310: if ('\0' != (sv = buf[*pos]))
1.52 kristaps 311: buf[(*pos)++] = '\0';
1.78 kristaps 312:
313: /*
314: * Now look up the word as a flag. Use temporary storage that
315: * we'll copy into the node's flags, if necessary.
316: */
1.1 kristaps 317:
1.68 kristaps 318: memset(&tmp, 0, sizeof(struct mdoc_argv));
1.78 kristaps 319:
1.1 kristaps 320: tmp.line = line;
321: tmp.pos = *pos;
1.78 kristaps 322: tmp.arg = MDOC_ARG_MAX;
1.1 kristaps 323:
1.78 kristaps 324: while (MDOC_ARG_MAX != (tmp.arg = *ap++))
325: if (0 == strcmp(p, mdoc_argnames[tmp.arg]))
326: break;
1.1 kristaps 327:
1.78 kristaps 328: if (MDOC_ARG_MAX == tmp.arg) {
329: /*
330: * The flag was not found.
331: * Restore saved zeroed byte and return as a word.
332: */
1.1 kristaps 333: if (sv)
334: buf[*pos - 1] = sv;
335: return(ARGV_WORD);
336: }
337:
1.78 kristaps 338: /* Read to the next word (the argument). */
339:
1.1 kristaps 340: while (buf[*pos] && ' ' == buf[*pos])
341: (*pos)++;
342:
1.79 kristaps 343: switch (argvflags[tmp.arg]) {
344: case (ARGV_SINGLE):
1.85 schwarze 345: if ( ! argv_single(mdoc, line, &tmp, pos, buf))
1.79 kristaps 346: return(ARGV_ERROR);
347: break;
348: case (ARGV_MULTI):
1.85 schwarze 349: if ( ! argv_multi(mdoc, line, &tmp, pos, buf))
1.79 kristaps 350: return(ARGV_ERROR);
351: break;
352: case (ARGV_NONE):
353: break;
354: }
1.1 kristaps 355:
1.31 kristaps 356: if (NULL == (arg = *v))
1.32 kristaps 357: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 358:
359: arg->argc++;
1.31 kristaps 360: arg->argv = mandoc_realloc
361: (arg->argv, arg->argc * sizeof(struct mdoc_argv));
1.2 kristaps 362:
1.68 kristaps 363: memcpy(&arg->argv[(int)arg->argc - 1],
1.1 kristaps 364: &tmp, sizeof(struct mdoc_argv));
365:
366: return(ARGV_ARG);
367: }
368:
369: void
370: mdoc_argv_free(struct mdoc_arg *p)
371: {
1.37 kristaps 372: int i;
1.1 kristaps 373:
374: if (NULL == p)
375: return;
376:
377: if (p->refcnt) {
378: --(p->refcnt);
379: if (p->refcnt)
380: return;
381: }
382: assert(p->argc);
383:
1.37 kristaps 384: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 385: argn_free(p, i);
1.37 kristaps 386:
387: free(p->argv);
388: free(p);
389: }
390:
1.71 kristaps 391: static void
392: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 393: {
1.58 kristaps 394: struct mdoc_argv *arg;
1.37 kristaps 395: int j;
1.58 kristaps 396:
397: arg = &p->argv[iarg];
1.1 kristaps 398:
1.37 kristaps 399: if (arg->sz && arg->value) {
400: for (j = (int)arg->sz - 1; j >= 0; j--)
401: free(arg->value[j]);
402: free(arg->value);
1.1 kristaps 403: }
404:
1.37 kristaps 405: for (--p->argc; iarg < (int)p->argc; iarg++)
406: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 407: }
408:
1.39 kristaps 409: enum margserr
1.85 schwarze 410: mdoc_zargs(struct mdoc *mdoc, int line, int *pos, char *buf, char **v)
1.19 kristaps 411: {
412:
1.85 schwarze 413: return(args(mdoc, line, pos, buf, ARGSFL_NONE, v));
1.19 kristaps 414: }
415:
1.39 kristaps 416: enum margserr
1.85 schwarze 417: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.35 kristaps 418: char *buf, enum mdoct tok, char **v)
1.1 kristaps 419: {
1.76 kristaps 420: enum argsflag fl;
1.1 kristaps 421: struct mdoc_node *n;
422:
1.78 kristaps 423: fl = mdocargs[tok].flags;
1.1 kristaps 424:
1.17 kristaps 425: if (MDOC_It != tok)
1.85 schwarze 426: return(args(mdoc, line, pos, buf, fl, v));
1.17 kristaps 427:
1.51 kristaps 428: /*
429: * We know that we're in an `It', so it's reasonable to expect
430: * us to be sitting in a `Bl'. Someday this may not be the case
431: * (if we allow random `It's sitting out there), so provide a
432: * safe fall-back into the default behaviour.
433: */
434:
1.85 schwarze 435: for (n = mdoc->last; n; n = n->parent)
1.50 kristaps 436: if (MDOC_Bl == n->tok)
1.76 kristaps 437: if (LIST_column == n->norm->Bl.type) {
438: fl = ARGSFL_TABSEP;
439: break;
440: }
1.1 kristaps 441:
1.85 schwarze 442: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 443: }
444:
1.39 kristaps 445: static enum margserr
1.85 schwarze 446: args(struct mdoc *mdoc, int line, int *pos,
1.76 kristaps 447: char *buf, enum argsflag fl, char **v)
1.1 kristaps 448: {
1.52 kristaps 449: char *p, *pp;
1.87 ! schwarze 450: int pairs;
1.52 kristaps 451: enum margserr rc;
1.1 kristaps 452:
1.46 kristaps 453: if ('\0' == buf[*pos]) {
1.85 schwarze 454: if (MDOC_PPHRASE & mdoc->flags)
1.46 kristaps 455: return(ARGS_EOLN);
456: /*
457: * If we're not in a partial phrase and the flag for
458: * being a phrase literal is still set, the punctuation
459: * is unterminated.
460: */
1.85 schwarze 461: if (MDOC_PHRASELIT & mdoc->flags)
462: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
1.46 kristaps 463:
1.85 schwarze 464: mdoc->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 465: return(ARGS_EOLN);
1.46 kristaps 466: }
1.1 kristaps 467:
1.64 kristaps 468: *v = &buf[*pos];
1.17 kristaps 469:
1.76 kristaps 470: if (ARGSFL_DELIM == fl)
1.77 kristaps 471: if (args_checkpunct(buf, *pos))
1.76 kristaps 472: return(ARGS_PUNCT);
1.1 kristaps 473:
1.17 kristaps 474: /*
475: * First handle TABSEP items, restricted to `Bl -column'. This
476: * ignores conventional token parsing and instead uses tabs or
477: * `Ta' macros to separate phrases. Phrases are parsed again
478: * for arguments at a later phase.
479: */
1.1 kristaps 480:
1.76 kristaps 481: if (ARGSFL_TABSEP == fl) {
1.20 kristaps 482: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 483: p = strchr(*v, '\t');
1.46 kristaps 484: pp = NULL;
1.17 kristaps 485:
486: /* Scan ahead to unescaped `Ta'. */
1.85 schwarze 487: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 488: for (pp = *v; ; pp++) {
489: if (NULL == (pp = strstr(pp, "Ta")))
490: break;
491: if (pp > *v && ' ' != *(pp - 1))
492: continue;
1.51 kristaps 493: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 494: break;
495: }
1.17 kristaps 496:
1.42 kristaps 497: /* By default, assume a phrase. */
498: rc = ARGS_PHRASE;
499:
1.21 kristaps 500: /*
501: * Adjust new-buffer position to be beyond delimiter
502: * mark (e.g., Ta -> end + 2).
503: */
1.20 kristaps 504: if (p && pp) {
505: *pos += pp < p ? 2 : 1;
1.42 kristaps 506: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 507: p = pp < p ? pp : p;
508: } else if (p && ! pp) {
1.42 kristaps 509: rc = ARGS_PPHRASE;
1.20 kristaps 510: *pos += 1;
511: } else if (pp && ! p) {
1.17 kristaps 512: p = pp;
1.20 kristaps 513: *pos += 2;
1.44 kristaps 514: } else {
515: rc = ARGS_PEND;
1.20 kristaps 516: p = strchr(*v, 0);
1.44 kristaps 517: }
1.20 kristaps 518:
1.21 kristaps 519: /* Whitespace check for eoln case... */
1.76 kristaps 520: if ('\0' == *p && ' ' == *(p - 1))
1.85 schwarze 521: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
1.20 kristaps 522:
523: *pos += (int)(p - *v);
1.17 kristaps 524:
525: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 526: pp = p - 1;
527: while (pp > *v && ' ' == *pp) {
528: if (pp > *v && '\\' == *(pp - 1))
529: break;
530: pp--;
1.17 kristaps 531: }
1.20 kristaps 532: *(pp + 1) = 0;
1.17 kristaps 533:
1.20 kristaps 534: /* Strip delimiter's proceeding whitespace. */
535: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
536: /* Skip ahead. */ ;
1.1 kristaps 537:
1.41 kristaps 538: return(rc);
1.17 kristaps 539: }
1.1 kristaps 540:
1.17 kristaps 541: /*
542: * Process a quoted literal. A quote begins with a double-quote
543: * and ends with a double-quote NOT preceded by a double-quote.
1.87 ! schwarze 544: * Null-terminate the literal in place.
! 545: * Collapse pairs of quotes inside quoted literals.
1.17 kristaps 546: * Whitespace is NOT involved in literal termination.
547: */
1.1 kristaps 548:
1.85 schwarze 549: if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
550: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 551: *v = &buf[++(*pos)];
552:
1.85 schwarze 553: if (MDOC_PPHRASE & mdoc->flags)
554: mdoc->flags |= MDOC_PHRASELIT;
1.1 kristaps 555:
1.87 ! schwarze 556: pairs = 0;
1.17 kristaps 557: for ( ; buf[*pos]; (*pos)++) {
1.87 ! schwarze 558: /* Move following text left after quoted quotes. */
! 559: if (pairs)
! 560: buf[*pos - pairs] = buf[*pos];
1.17 kristaps 561: if ('\"' != buf[*pos])
562: continue;
1.87 ! schwarze 563: /* Unquoted quotes end quoted args. */
1.17 kristaps 564: if ('\"' != buf[*pos + 1])
565: break;
1.87 ! schwarze 566: /* Quoted quotes collapse. */
! 567: pairs++;
1.17 kristaps 568: (*pos)++;
569: }
1.87 ! schwarze 570: if (pairs)
! 571: buf[*pos - pairs] = '\0';
1.1 kristaps 572:
1.46 kristaps 573: if ('\0' == buf[*pos]) {
1.85 schwarze 574: if (MDOC_PPHRASE & mdoc->flags)
1.23 kristaps 575: return(ARGS_QWORD);
1.85 schwarze 576: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
1.18 kristaps 577: return(ARGS_QWORD);
1.17 kristaps 578: }
1.1 kristaps 579:
1.85 schwarze 580: mdoc->flags &= ~MDOC_PHRASELIT;
1.46 kristaps 581: buf[(*pos)++] = '\0';
1.1 kristaps 582:
1.46 kristaps 583: if ('\0' == buf[*pos])
1.17 kristaps 584: return(ARGS_QWORD);
1.1 kristaps 585:
1.17 kristaps 586: while (' ' == buf[*pos])
587: (*pos)++;
1.1 kristaps 588:
1.76 kristaps 589: if ('\0' == buf[*pos])
1.85 schwarze 590: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 591:
1.17 kristaps 592: return(ARGS_QWORD);
1.1 kristaps 593: }
594:
1.75 kristaps 595: p = &buf[*pos];
1.85 schwarze 596: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 597:
1.17 kristaps 598: return(ARGS_WORD);
1.64 kristaps 599: }
600:
601: /*
602: * Check if the string consists only of space-separated closing
1.65 kristaps 603: * delimiters. This is a bit of a dance: the first must be a close
604: * delimiter, but it may be followed by middle delimiters. Arbitrary
605: * whitespace may separate these tokens.
1.64 kristaps 606: */
607: static int
1.77 kristaps 608: args_checkpunct(const char *buf, int i)
1.64 kristaps 609: {
1.69 kristaps 610: int j;
611: char dbuf[DELIMSZ];
1.64 kristaps 612: enum mdelim d;
613:
1.65 kristaps 614: /* First token must be a close-delimiter. */
615:
1.69 kristaps 616: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
617: dbuf[j] = buf[i];
1.65 kristaps 618:
619: if (DELIMSZ == j)
620: return(0);
621:
1.69 kristaps 622: dbuf[j] = '\0';
1.70 kristaps 623: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.64 kristaps 624: return(0);
625:
1.69 kristaps 626: while (' ' == buf[i])
1.65 kristaps 627: i++;
628:
629: /* Remaining must NOT be open/none. */
630:
1.69 kristaps 631: while (buf[i]) {
1.65 kristaps 632: j = 0;
1.69 kristaps 633: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
634: dbuf[j++] = buf[i++];
1.65 kristaps 635:
636: if (DELIMSZ == j)
637: return(0);
638:
1.69 kristaps 639: dbuf[j] = '\0';
1.70 kristaps 640: d = mdoc_isdelim(dbuf);
1.64 kristaps 641: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 642: return(0);
643:
1.69 kristaps 644: while (' ' == buf[i])
1.64 kristaps 645: i++;
646: }
1.69 kristaps 647:
648: return('\0' == buf[i]);
1.1 kristaps 649: }
650:
651: static int
1.85 schwarze 652: argv_multi(struct mdoc *mdoc, int line,
1.1 kristaps 653: struct mdoc_argv *v, int *pos, char *buf)
654: {
1.43 kristaps 655: enum margserr ac;
1.1 kristaps 656: char *p;
657:
658: for (v->sz = 0; ; v->sz++) {
659: if ('-' == buf[*pos])
660: break;
1.85 schwarze 661: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 662: if (ARGS_ERROR == ac)
1.1 kristaps 663: return(0);
1.43 kristaps 664: else if (ARGS_EOLN == ac)
1.1 kristaps 665: break;
666:
1.31 kristaps 667: if (0 == v->sz % MULTI_STEP)
668: v->value = mandoc_realloc(v->value,
1.1 kristaps 669: (v->sz + MULTI_STEP) * sizeof(char *));
1.31 kristaps 670:
671: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 672: }
673:
1.7 kristaps 674: return(1);
1.1 kristaps 675: }
676:
677: static int
1.85 schwarze 678: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 679: struct mdoc_argv *v, int *pos, char *buf)
680: {
1.43 kristaps 681: enum margserr ac;
1.1 kristaps 682: char *p;
683:
1.85 schwarze 684: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 685: if (ARGS_ERROR == ac)
1.1 kristaps 686: return(0);
1.43 kristaps 687: if (ARGS_EOLN == ac)
1.1 kristaps 688: return(1);
689:
690: v->sz = 1;
1.31 kristaps 691: v->value = mandoc_malloc(sizeof(char *));
692: v->value[0] = mandoc_strdup(p);
1.1 kristaps 693:
694: return(1);
695: }
CVSweb