Annotation of mandoc/mdoc_argv.c, Revision 1.90
1.90 ! schwarze 1: /* $Id: mdoc_argv.c,v 1.89 2013/12/25 00:50:05 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"
1.90 ! schwarze 31: #include "mandoc_aux.h"
1.1 kristaps 32: #include "libmdoc.h"
1.31 kristaps 33: #include "libmandoc.h"
1.1 kristaps 34:
1.68 kristaps 35: #define MULTI_STEP 5 /* pre-allocate argument values */
1.76 kristaps 36: #define DELIMSZ 6 /* max possible size of a delimiter */
37:
38: enum argsflag {
39: ARGSFL_NONE = 0,
40: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
41: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
42: };
43:
44: enum argvflag {
45: ARGV_NONE, /* no args to flag (e.g., -split) */
46: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
1.83 schwarze 47: ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
1.76 kristaps 48: };
1.1 kristaps 49:
1.78 kristaps 50: struct mdocarg {
51: enum argsflag flags;
52: const enum mdocargt *argvs;
53: };
54:
1.80 kristaps 55: static void argn_free(struct mdoc_arg *, int);
1.39 kristaps 56: static enum margserr args(struct mdoc *, int, int *,
1.76 kristaps 57: char *, enum argsflag, char **);
1.77 kristaps 58: static int args_checkpunct(const char *, int);
1.80 kristaps 59: static int argv_multi(struct mdoc *, int,
1.1 kristaps 60: struct mdoc_argv *, int *, char *);
1.80 kristaps 61: static int argv_single(struct mdoc *, int,
1.1 kristaps 62: struct mdoc_argv *, int *, char *);
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 */
1.83 schwarze 71: ARGV_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 */
1.83 schwarze 83: ARGV_SINGLE, /* MDOC_Width */
1.1 kristaps 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:
1.73 kristaps 93: static const enum mdocargt args_Ex[] = {
94: MDOC_Std,
95: MDOC_ARG_MAX
96: };
97:
98: static const enum mdocargt args_An[] = {
99: MDOC_Split,
100: MDOC_Nosplit,
101: MDOC_ARG_MAX
102: };
103:
104: static const enum mdocargt args_Bd[] = {
105: MDOC_Ragged,
106: MDOC_Unfilled,
107: MDOC_Filled,
108: MDOC_Literal,
109: MDOC_File,
110: MDOC_Offset,
111: MDOC_Compact,
112: MDOC_Centred,
113: MDOC_ARG_MAX
114: };
115:
116: static const enum mdocargt args_Bf[] = {
117: MDOC_Emphasis,
118: MDOC_Literal,
119: MDOC_Symbolic,
120: MDOC_ARG_MAX
121: };
122:
123: static const enum mdocargt args_Bk[] = {
124: MDOC_Words,
125: MDOC_ARG_MAX
126: };
127:
128: static const enum mdocargt args_Bl[] = {
129: MDOC_Bullet,
130: MDOC_Dash,
131: MDOC_Hyphen,
132: MDOC_Item,
133: MDOC_Enum,
134: MDOC_Tag,
135: MDOC_Diag,
136: MDOC_Hang,
137: MDOC_Ohang,
138: MDOC_Inset,
139: MDOC_Column,
140: MDOC_Width,
141: MDOC_Offset,
142: MDOC_Compact,
143: MDOC_Nested,
144: MDOC_ARG_MAX
145: };
146:
1.78 kristaps 147: static const struct mdocarg mdocargs[MDOC_MAX] = {
1.84 schwarze 148: { ARGSFL_DELIM, NULL }, /* Ap */
1.78 kristaps 149: { ARGSFL_NONE, NULL }, /* Dd */
150: { ARGSFL_NONE, NULL }, /* Dt */
151: { ARGSFL_NONE, NULL }, /* Os */
152: { ARGSFL_NONE, NULL }, /* Sh */
153: { ARGSFL_NONE, NULL }, /* Ss */
154: { ARGSFL_NONE, NULL }, /* Pp */
155: { ARGSFL_DELIM, NULL }, /* D1 */
156: { ARGSFL_DELIM, NULL }, /* Dl */
157: { ARGSFL_NONE, args_Bd }, /* Bd */
158: { ARGSFL_NONE, NULL }, /* Ed */
159: { ARGSFL_NONE, args_Bl }, /* Bl */
160: { ARGSFL_NONE, NULL }, /* El */
161: { ARGSFL_NONE, NULL }, /* It */
162: { ARGSFL_DELIM, NULL }, /* Ad */
163: { ARGSFL_DELIM, args_An }, /* An */
164: { ARGSFL_DELIM, NULL }, /* Ar */
1.86 schwarze 165: { ARGSFL_DELIM, NULL }, /* Cd */
1.78 kristaps 166: { ARGSFL_DELIM, NULL }, /* Cm */
167: { ARGSFL_DELIM, NULL }, /* Dv */
168: { ARGSFL_DELIM, NULL }, /* Er */
169: { ARGSFL_DELIM, NULL }, /* Ev */
170: { ARGSFL_NONE, args_Ex }, /* Ex */
171: { ARGSFL_DELIM, NULL }, /* Fa */
172: { ARGSFL_NONE, NULL }, /* Fd */
173: { ARGSFL_DELIM, NULL }, /* Fl */
174: { ARGSFL_DELIM, NULL }, /* Fn */
175: { ARGSFL_DELIM, NULL }, /* Ft */
176: { ARGSFL_DELIM, NULL }, /* Ic */
1.84 schwarze 177: { ARGSFL_DELIM, NULL }, /* In */
1.78 kristaps 178: { ARGSFL_DELIM, NULL }, /* Li */
179: { ARGSFL_NONE, NULL }, /* Nd */
180: { ARGSFL_DELIM, NULL }, /* Nm */
181: { ARGSFL_DELIM, NULL }, /* Op */
182: { ARGSFL_NONE, NULL }, /* Ot */
183: { ARGSFL_DELIM, NULL }, /* Pa */
184: { ARGSFL_NONE, args_Ex }, /* Rv */
185: { ARGSFL_DELIM, NULL }, /* St */
186: { ARGSFL_DELIM, NULL }, /* Va */
187: { ARGSFL_DELIM, NULL }, /* Vt */
188: { ARGSFL_DELIM, NULL }, /* Xr */
189: { ARGSFL_NONE, NULL }, /* %A */
190: { ARGSFL_NONE, NULL }, /* %B */
191: { ARGSFL_NONE, NULL }, /* %D */
192: { ARGSFL_NONE, NULL }, /* %I */
193: { ARGSFL_NONE, NULL }, /* %J */
194: { ARGSFL_NONE, NULL }, /* %N */
195: { ARGSFL_NONE, NULL }, /* %O */
196: { ARGSFL_NONE, NULL }, /* %P */
197: { ARGSFL_NONE, NULL }, /* %R */
198: { ARGSFL_NONE, NULL }, /* %T */
199: { ARGSFL_NONE, NULL }, /* %V */
200: { ARGSFL_DELIM, NULL }, /* Ac */
201: { ARGSFL_NONE, NULL }, /* Ao */
202: { ARGSFL_DELIM, NULL }, /* Aq */
203: { ARGSFL_DELIM, NULL }, /* At */
204: { ARGSFL_DELIM, NULL }, /* Bc */
205: { ARGSFL_NONE, args_Bf }, /* Bf */
206: { ARGSFL_NONE, NULL }, /* Bo */
207: { ARGSFL_DELIM, NULL }, /* Bq */
208: { ARGSFL_DELIM, NULL }, /* Bsx */
209: { ARGSFL_DELIM, NULL }, /* Bx */
210: { ARGSFL_NONE, NULL }, /* Db */
211: { ARGSFL_DELIM, NULL }, /* Dc */
212: { ARGSFL_NONE, NULL }, /* Do */
213: { ARGSFL_DELIM, NULL }, /* Dq */
214: { ARGSFL_DELIM, NULL }, /* Ec */
215: { ARGSFL_NONE, NULL }, /* Ef */
216: { ARGSFL_DELIM, NULL }, /* Em */
217: { ARGSFL_NONE, NULL }, /* Eo */
218: { ARGSFL_DELIM, NULL }, /* Fx */
219: { ARGSFL_DELIM, NULL }, /* Ms */
220: { ARGSFL_DELIM, NULL }, /* No */
221: { ARGSFL_DELIM, NULL }, /* Ns */
222: { ARGSFL_DELIM, NULL }, /* Nx */
223: { ARGSFL_DELIM, NULL }, /* Ox */
224: { ARGSFL_DELIM, NULL }, /* Pc */
225: { ARGSFL_DELIM, NULL }, /* Pf */
226: { ARGSFL_NONE, NULL }, /* Po */
227: { ARGSFL_DELIM, NULL }, /* Pq */
228: { ARGSFL_DELIM, NULL }, /* Qc */
229: { ARGSFL_DELIM, NULL }, /* Ql */
230: { ARGSFL_NONE, NULL }, /* Qo */
231: { ARGSFL_DELIM, NULL }, /* Qq */
232: { ARGSFL_NONE, NULL }, /* Re */
233: { ARGSFL_NONE, NULL }, /* Rs */
234: { ARGSFL_DELIM, NULL }, /* Sc */
235: { ARGSFL_NONE, NULL }, /* So */
236: { ARGSFL_DELIM, NULL }, /* Sq */
237: { ARGSFL_NONE, NULL }, /* Sm */
238: { ARGSFL_DELIM, NULL }, /* Sx */
239: { ARGSFL_DELIM, NULL }, /* Sy */
240: { ARGSFL_DELIM, NULL }, /* Tn */
241: { ARGSFL_DELIM, NULL }, /* Ux */
242: { ARGSFL_DELIM, NULL }, /* Xc */
243: { ARGSFL_NONE, NULL }, /* Xo */
244: { ARGSFL_NONE, NULL }, /* Fo */
1.86 schwarze 245: { ARGSFL_DELIM, NULL }, /* Fc */
1.78 kristaps 246: { ARGSFL_NONE, NULL }, /* Oo */
247: { ARGSFL_DELIM, NULL }, /* Oc */
248: { ARGSFL_NONE, args_Bk }, /* Bk */
249: { ARGSFL_NONE, NULL }, /* Ek */
250: { ARGSFL_NONE, NULL }, /* Bt */
251: { ARGSFL_NONE, NULL }, /* Hf */
252: { ARGSFL_NONE, NULL }, /* Fr */
253: { ARGSFL_NONE, NULL }, /* Ud */
1.86 schwarze 254: { ARGSFL_DELIM, NULL }, /* Lb */
1.78 kristaps 255: { ARGSFL_NONE, NULL }, /* Lp */
256: { ARGSFL_DELIM, NULL }, /* Lk */
257: { ARGSFL_DELIM, NULL }, /* Mt */
258: { ARGSFL_DELIM, NULL }, /* Brq */
259: { ARGSFL_NONE, NULL }, /* Bro */
260: { ARGSFL_DELIM, NULL }, /* Brc */
261: { ARGSFL_NONE, NULL }, /* %C */
262: { ARGSFL_NONE, NULL }, /* Es */
263: { ARGSFL_NONE, NULL }, /* En */
1.88 schwarze 264: { ARGSFL_DELIM, NULL }, /* Dx */
1.78 kristaps 265: { ARGSFL_NONE, NULL }, /* %Q */
266: { ARGSFL_NONE, NULL }, /* br */
267: { ARGSFL_NONE, NULL }, /* sp */
268: { ARGSFL_NONE, NULL }, /* %U */
269: { ARGSFL_NONE, NULL }, /* Ta */
270: };
271:
272:
1.1 kristaps 273: /*
274: * Parse an argument from line text. This comes in the form of -key
275: * [value0...], which may either have a single mandatory value, at least
276: * one mandatory value, an optional single value, or no value.
277: */
1.40 kristaps 278: enum margverr
1.85 schwarze 279: mdoc_argv(struct mdoc *mdoc, int line, enum mdoct tok,
1.1 kristaps 280: struct mdoc_arg **v, int *pos, char *buf)
281: {
282: char *p, sv;
283: struct mdoc_argv tmp;
284: struct mdoc_arg *arg;
1.78 kristaps 285: const enum mdocargt *ap;
1.1 kristaps 286:
1.52 kristaps 287: if ('\0' == buf[*pos])
1.1 kristaps 288: return(ARGV_EOLN);
1.78 kristaps 289: else if (NULL == (ap = mdocargs[tok].argvs))
290: return(ARGV_WORD);
1.80 kristaps 291: else if ('-' != buf[*pos])
292: return(ARGV_WORD);
1.1 kristaps 293:
1.78 kristaps 294: /* Seek to the first unescaped space. */
1.1 kristaps 295:
296: p = &buf[++(*pos)];
297:
298: assert(*pos > 0);
299:
1.78 kristaps 300: for ( ; buf[*pos] ; (*pos)++)
301: if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
302: break;
1.1 kristaps 303:
1.78 kristaps 304: /*
305: * We want to nil-terminate the word to look it up (it's easier
306: * that way). But we may not have a flag, in which case we need
307: * to restore the line as-is. So keep around the stray byte,
308: * which we'll reset upon exiting (if necessary).
309: */
1.1 kristaps 310:
1.78 kristaps 311: if ('\0' != (sv = buf[*pos]))
1.52 kristaps 312: buf[(*pos)++] = '\0';
1.78 kristaps 313:
314: /*
315: * Now look up the word as a flag. Use temporary storage that
316: * we'll copy into the node's flags, if necessary.
317: */
1.1 kristaps 318:
1.68 kristaps 319: memset(&tmp, 0, sizeof(struct mdoc_argv));
1.78 kristaps 320:
1.1 kristaps 321: tmp.line = line;
322: tmp.pos = *pos;
1.78 kristaps 323: tmp.arg = MDOC_ARG_MAX;
1.1 kristaps 324:
1.78 kristaps 325: while (MDOC_ARG_MAX != (tmp.arg = *ap++))
326: if (0 == strcmp(p, mdoc_argnames[tmp.arg]))
327: break;
1.1 kristaps 328:
1.78 kristaps 329: if (MDOC_ARG_MAX == tmp.arg) {
330: /*
331: * The flag was not found.
332: * Restore saved zeroed byte and return as a word.
333: */
1.1 kristaps 334: if (sv)
335: buf[*pos - 1] = sv;
336: return(ARGV_WORD);
337: }
338:
1.78 kristaps 339: /* Read to the next word (the argument). */
340:
1.1 kristaps 341: while (buf[*pos] && ' ' == buf[*pos])
342: (*pos)++;
343:
1.79 kristaps 344: switch (argvflags[tmp.arg]) {
345: case (ARGV_SINGLE):
1.85 schwarze 346: if ( ! argv_single(mdoc, line, &tmp, pos, buf))
1.79 kristaps 347: return(ARGV_ERROR);
348: break;
349: case (ARGV_MULTI):
1.85 schwarze 350: if ( ! argv_multi(mdoc, line, &tmp, pos, buf))
1.79 kristaps 351: return(ARGV_ERROR);
352: break;
353: case (ARGV_NONE):
354: break;
355: }
1.1 kristaps 356:
1.31 kristaps 357: if (NULL == (arg = *v))
1.32 kristaps 358: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 359:
360: arg->argc++;
1.31 kristaps 361: arg->argv = mandoc_realloc
362: (arg->argv, arg->argc * sizeof(struct mdoc_argv));
1.2 kristaps 363:
1.68 kristaps 364: memcpy(&arg->argv[(int)arg->argc - 1],
1.1 kristaps 365: &tmp, sizeof(struct mdoc_argv));
366:
367: return(ARGV_ARG);
368: }
369:
370: void
371: mdoc_argv_free(struct mdoc_arg *p)
372: {
1.37 kristaps 373: int i;
1.1 kristaps 374:
375: if (NULL == p)
376: return;
377:
378: if (p->refcnt) {
379: --(p->refcnt);
380: if (p->refcnt)
381: return;
382: }
383: assert(p->argc);
384:
1.37 kristaps 385: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 386: argn_free(p, i);
1.37 kristaps 387:
388: free(p->argv);
389: free(p);
390: }
391:
1.71 kristaps 392: static void
393: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 394: {
1.58 kristaps 395: struct mdoc_argv *arg;
1.37 kristaps 396: int j;
1.58 kristaps 397:
398: arg = &p->argv[iarg];
1.1 kristaps 399:
1.37 kristaps 400: if (arg->sz && arg->value) {
401: for (j = (int)arg->sz - 1; j >= 0; j--)
402: free(arg->value[j]);
403: free(arg->value);
1.1 kristaps 404: }
405:
1.37 kristaps 406: for (--p->argc; iarg < (int)p->argc; iarg++)
407: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 408: }
409:
1.39 kristaps 410: enum margserr
1.85 schwarze 411: mdoc_zargs(struct mdoc *mdoc, int line, int *pos, char *buf, char **v)
1.19 kristaps 412: {
413:
1.85 schwarze 414: return(args(mdoc, line, pos, buf, ARGSFL_NONE, v));
1.19 kristaps 415: }
416:
1.39 kristaps 417: enum margserr
1.85 schwarze 418: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.35 kristaps 419: char *buf, enum mdoct tok, char **v)
1.1 kristaps 420: {
1.76 kristaps 421: enum argsflag fl;
1.1 kristaps 422: struct mdoc_node *n;
423:
1.78 kristaps 424: fl = mdocargs[tok].flags;
1.1 kristaps 425:
1.17 kristaps 426: if (MDOC_It != tok)
1.85 schwarze 427: return(args(mdoc, line, pos, buf, fl, v));
1.17 kristaps 428:
1.51 kristaps 429: /*
430: * We know that we're in an `It', so it's reasonable to expect
431: * us to be sitting in a `Bl'. Someday this may not be the case
432: * (if we allow random `It's sitting out there), so provide a
433: * safe fall-back into the default behaviour.
434: */
435:
1.85 schwarze 436: for (n = mdoc->last; n; n = n->parent)
1.50 kristaps 437: if (MDOC_Bl == n->tok)
1.76 kristaps 438: if (LIST_column == n->norm->Bl.type) {
439: fl = ARGSFL_TABSEP;
440: break;
441: }
1.1 kristaps 442:
1.85 schwarze 443: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 444: }
445:
1.39 kristaps 446: static enum margserr
1.85 schwarze 447: args(struct mdoc *mdoc, int line, int *pos,
1.76 kristaps 448: char *buf, enum argsflag fl, char **v)
1.1 kristaps 449: {
1.52 kristaps 450: char *p, *pp;
1.87 schwarze 451: int pairs;
1.52 kristaps 452: enum margserr rc;
1.1 kristaps 453:
1.46 kristaps 454: if ('\0' == buf[*pos]) {
1.85 schwarze 455: if (MDOC_PPHRASE & mdoc->flags)
1.46 kristaps 456: return(ARGS_EOLN);
457: /*
458: * If we're not in a partial phrase and the flag for
459: * being a phrase literal is still set, the punctuation
460: * is unterminated.
461: */
1.85 schwarze 462: if (MDOC_PHRASELIT & mdoc->flags)
463: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
1.46 kristaps 464:
1.85 schwarze 465: mdoc->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 466: return(ARGS_EOLN);
1.46 kristaps 467: }
1.1 kristaps 468:
1.64 kristaps 469: *v = &buf[*pos];
1.17 kristaps 470:
1.76 kristaps 471: if (ARGSFL_DELIM == fl)
1.77 kristaps 472: if (args_checkpunct(buf, *pos))
1.76 kristaps 473: return(ARGS_PUNCT);
1.1 kristaps 474:
1.17 kristaps 475: /*
476: * First handle TABSEP items, restricted to `Bl -column'. This
477: * ignores conventional token parsing and instead uses tabs or
478: * `Ta' macros to separate phrases. Phrases are parsed again
479: * for arguments at a later phase.
480: */
1.1 kristaps 481:
1.76 kristaps 482: if (ARGSFL_TABSEP == fl) {
1.20 kristaps 483: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 484: p = strchr(*v, '\t');
1.46 kristaps 485: pp = NULL;
1.17 kristaps 486:
487: /* Scan ahead to unescaped `Ta'. */
1.85 schwarze 488: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 489: for (pp = *v; ; pp++) {
490: if (NULL == (pp = strstr(pp, "Ta")))
491: break;
492: if (pp > *v && ' ' != *(pp - 1))
493: continue;
1.51 kristaps 494: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 495: break;
496: }
1.17 kristaps 497:
1.42 kristaps 498: /* By default, assume a phrase. */
499: rc = ARGS_PHRASE;
500:
1.21 kristaps 501: /*
502: * Adjust new-buffer position to be beyond delimiter
503: * mark (e.g., Ta -> end + 2).
504: */
1.20 kristaps 505: if (p && pp) {
506: *pos += pp < p ? 2 : 1;
1.42 kristaps 507: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 508: p = pp < p ? pp : p;
509: } else if (p && ! pp) {
1.42 kristaps 510: rc = ARGS_PPHRASE;
1.20 kristaps 511: *pos += 1;
512: } else if (pp && ! p) {
1.17 kristaps 513: p = pp;
1.20 kristaps 514: *pos += 2;
1.44 kristaps 515: } else {
516: rc = ARGS_PEND;
1.20 kristaps 517: p = strchr(*v, 0);
1.44 kristaps 518: }
1.20 kristaps 519:
1.21 kristaps 520: /* Whitespace check for eoln case... */
1.76 kristaps 521: if ('\0' == *p && ' ' == *(p - 1))
1.85 schwarze 522: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
1.20 kristaps 523:
524: *pos += (int)(p - *v);
1.17 kristaps 525:
526: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 527: pp = p - 1;
528: while (pp > *v && ' ' == *pp) {
529: if (pp > *v && '\\' == *(pp - 1))
530: break;
531: pp--;
1.17 kristaps 532: }
1.20 kristaps 533: *(pp + 1) = 0;
1.17 kristaps 534:
1.20 kristaps 535: /* Strip delimiter's proceeding whitespace. */
536: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
537: /* Skip ahead. */ ;
1.1 kristaps 538:
1.41 kristaps 539: return(rc);
1.89 schwarze 540: }
1.1 kristaps 541:
1.89 schwarze 542: /*
1.17 kristaps 543: * Process a quoted literal. A quote begins with a double-quote
544: * and ends with a double-quote NOT preceded by a double-quote.
1.89 schwarze 545: * NUL-terminate the literal in place.
1.87 schwarze 546: * Collapse pairs of quotes inside quoted literals.
1.17 kristaps 547: * Whitespace is NOT involved in literal termination.
548: */
1.1 kristaps 549:
1.85 schwarze 550: if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
551: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 552: *v = &buf[++(*pos)];
553:
1.85 schwarze 554: if (MDOC_PPHRASE & mdoc->flags)
555: mdoc->flags |= MDOC_PHRASELIT;
1.1 kristaps 556:
1.87 schwarze 557: pairs = 0;
1.17 kristaps 558: for ( ; buf[*pos]; (*pos)++) {
1.87 schwarze 559: /* Move following text left after quoted quotes. */
560: if (pairs)
561: buf[*pos - pairs] = buf[*pos];
1.17 kristaps 562: if ('\"' != buf[*pos])
563: continue;
1.87 schwarze 564: /* Unquoted quotes end quoted args. */
1.17 kristaps 565: if ('\"' != buf[*pos + 1])
566: break;
1.87 schwarze 567: /* Quoted quotes collapse. */
568: pairs++;
1.17 kristaps 569: (*pos)++;
570: }
1.87 schwarze 571: if (pairs)
572: buf[*pos - pairs] = '\0';
1.1 kristaps 573:
1.46 kristaps 574: if ('\0' == buf[*pos]) {
1.85 schwarze 575: if (MDOC_PPHRASE & mdoc->flags)
1.23 kristaps 576: return(ARGS_QWORD);
1.85 schwarze 577: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
1.18 kristaps 578: return(ARGS_QWORD);
1.17 kristaps 579: }
1.1 kristaps 580:
1.85 schwarze 581: mdoc->flags &= ~MDOC_PHRASELIT;
1.46 kristaps 582: buf[(*pos)++] = '\0';
1.1 kristaps 583:
1.46 kristaps 584: if ('\0' == buf[*pos])
1.17 kristaps 585: return(ARGS_QWORD);
1.1 kristaps 586:
1.17 kristaps 587: while (' ' == buf[*pos])
588: (*pos)++;
1.1 kristaps 589:
1.76 kristaps 590: if ('\0' == buf[*pos])
1.85 schwarze 591: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 592:
1.17 kristaps 593: return(ARGS_QWORD);
1.1 kristaps 594: }
595:
1.75 kristaps 596: p = &buf[*pos];
1.85 schwarze 597: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 598:
1.17 kristaps 599: return(ARGS_WORD);
1.64 kristaps 600: }
601:
602: /*
603: * Check if the string consists only of space-separated closing
1.65 kristaps 604: * delimiters. This is a bit of a dance: the first must be a close
605: * delimiter, but it may be followed by middle delimiters. Arbitrary
606: * whitespace may separate these tokens.
1.64 kristaps 607: */
608: static int
1.77 kristaps 609: args_checkpunct(const char *buf, int i)
1.64 kristaps 610: {
1.69 kristaps 611: int j;
612: char dbuf[DELIMSZ];
1.64 kristaps 613: enum mdelim d;
614:
1.65 kristaps 615: /* First token must be a close-delimiter. */
616:
1.69 kristaps 617: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
618: dbuf[j] = buf[i];
1.65 kristaps 619:
620: if (DELIMSZ == j)
621: return(0);
622:
1.69 kristaps 623: dbuf[j] = '\0';
1.70 kristaps 624: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.64 kristaps 625: return(0);
626:
1.69 kristaps 627: while (' ' == buf[i])
1.65 kristaps 628: i++;
629:
630: /* Remaining must NOT be open/none. */
631:
1.69 kristaps 632: while (buf[i]) {
1.65 kristaps 633: j = 0;
1.69 kristaps 634: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
635: dbuf[j++] = buf[i++];
1.65 kristaps 636:
637: if (DELIMSZ == j)
638: return(0);
639:
1.69 kristaps 640: dbuf[j] = '\0';
1.70 kristaps 641: d = mdoc_isdelim(dbuf);
1.64 kristaps 642: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 643: return(0);
644:
1.69 kristaps 645: while (' ' == buf[i])
1.64 kristaps 646: i++;
647: }
1.69 kristaps 648:
649: return('\0' == buf[i]);
1.1 kristaps 650: }
651:
652: static int
1.85 schwarze 653: argv_multi(struct mdoc *mdoc, int line,
1.1 kristaps 654: struct mdoc_argv *v, int *pos, char *buf)
655: {
1.43 kristaps 656: enum margserr ac;
1.1 kristaps 657: char *p;
658:
659: for (v->sz = 0; ; v->sz++) {
660: if ('-' == buf[*pos])
661: break;
1.85 schwarze 662: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 663: if (ARGS_ERROR == ac)
1.1 kristaps 664: return(0);
1.43 kristaps 665: else if (ARGS_EOLN == ac)
1.1 kristaps 666: break;
667:
1.31 kristaps 668: if (0 == v->sz % MULTI_STEP)
669: v->value = mandoc_realloc(v->value,
1.1 kristaps 670: (v->sz + MULTI_STEP) * sizeof(char *));
1.31 kristaps 671:
672: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 673: }
674:
1.7 kristaps 675: return(1);
1.1 kristaps 676: }
677:
678: static int
1.85 schwarze 679: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 680: struct mdoc_argv *v, int *pos, char *buf)
681: {
1.43 kristaps 682: enum margserr ac;
1.1 kristaps 683: char *p;
684:
1.85 schwarze 685: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 686: if (ARGS_ERROR == ac)
1.1 kristaps 687: return(0);
1.43 kristaps 688: if (ARGS_EOLN == ac)
1.1 kristaps 689: return(1);
690:
691: v->sz = 1;
1.31 kristaps 692: v->value = mandoc_malloc(sizeof(char *));
693: v->value[0] = mandoc_strdup(p);
1.1 kristaps 694:
695: return(1);
696: }
CVSweb