Annotation of src/usr.bin/mandoc/mdoc_argv.c, Revision 1.46
1.46 ! schwarze 1: /* $Id: mdoc_argv.c,v 1.45 2013/12/22 14:06:29 schwarze Exp $ */
1.1 kristaps 2: /*
1.36 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.41 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
1.2 schwarze 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: *
1.2 schwarze 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: */
18: #include <sys/types.h>
19:
20: #include <assert.h>
21: #include <stdlib.h>
22: #include <stdio.h>
23: #include <string.h>
24:
1.37 schwarze 25: #include "mdoc.h"
1.29 schwarze 26: #include "mandoc.h"
1.1 kristaps 27: #include "libmdoc.h"
1.19 schwarze 28: #include "libmandoc.h"
1.1 kristaps 29:
1.36 schwarze 30: #define MULTI_STEP 5 /* pre-allocate argument values */
1.38 schwarze 31: #define DELIMSZ 6 /* max possible size of a delimiter */
32:
33: enum argsflag {
34: ARGSFL_NONE = 0,
35: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
36: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
37: };
38:
39: enum argvflag {
40: ARGV_NONE, /* no args to flag (e.g., -split) */
41: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
1.40 schwarze 42: ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
1.38 schwarze 43: };
1.1 kristaps 44:
1.39 schwarze 45: struct mdocarg {
46: enum argsflag flags;
47: const enum mdocargt *argvs;
48: };
49:
50: static void argn_free(struct mdoc_arg *, int);
1.24 schwarze 51: static enum margserr args(struct mdoc *, int, int *,
1.38 schwarze 52: char *, enum argsflag, char **);
53: static int args_checkpunct(const char *, int);
1.39 schwarze 54: static int argv_multi(struct mdoc *, int,
1.1 kristaps 55: struct mdoc_argv *, int *, char *);
1.39 schwarze 56: static int argv_single(struct mdoc *, int,
1.1 kristaps 57: struct mdoc_argv *, int *, char *);
58:
1.36 schwarze 59: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 60: ARGV_NONE, /* MDOC_Split */
61: ARGV_NONE, /* MDOC_Nosplit */
62: ARGV_NONE, /* MDOC_Ragged */
63: ARGV_NONE, /* MDOC_Unfilled */
64: ARGV_NONE, /* MDOC_Literal */
1.17 schwarze 65: ARGV_SINGLE, /* MDOC_File */
1.40 schwarze 66: ARGV_SINGLE, /* MDOC_Offset */
1.1 kristaps 67: ARGV_NONE, /* MDOC_Bullet */
68: ARGV_NONE, /* MDOC_Dash */
69: ARGV_NONE, /* MDOC_Hyphen */
70: ARGV_NONE, /* MDOC_Item */
71: ARGV_NONE, /* MDOC_Enum */
72: ARGV_NONE, /* MDOC_Tag */
73: ARGV_NONE, /* MDOC_Diag */
74: ARGV_NONE, /* MDOC_Hang */
75: ARGV_NONE, /* MDOC_Ohang */
76: ARGV_NONE, /* MDOC_Inset */
77: ARGV_MULTI, /* MDOC_Column */
78: ARGV_SINGLE, /* MDOC_Width */
79: ARGV_NONE, /* MDOC_Compact */
1.14 schwarze 80: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 81: ARGV_NONE, /* MDOC_Filled */
82: ARGV_NONE, /* MDOC_Words */
83: ARGV_NONE, /* MDOC_Emphasis */
84: ARGV_NONE, /* MDOC_Symbolic */
85: ARGV_NONE /* MDOC_Symbolic */
86: };
87:
1.37 schwarze 88: static const enum mdocargt args_Ex[] = {
89: MDOC_Std,
90: MDOC_ARG_MAX
91: };
92:
93: static const enum mdocargt args_An[] = {
94: MDOC_Split,
95: MDOC_Nosplit,
96: MDOC_ARG_MAX
97: };
98:
99: static const enum mdocargt args_Bd[] = {
100: MDOC_Ragged,
101: MDOC_Unfilled,
102: MDOC_Filled,
103: MDOC_Literal,
104: MDOC_File,
105: MDOC_Offset,
106: MDOC_Compact,
107: MDOC_Centred,
108: MDOC_ARG_MAX
109: };
110:
111: static const enum mdocargt args_Bf[] = {
112: MDOC_Emphasis,
113: MDOC_Literal,
114: MDOC_Symbolic,
115: MDOC_ARG_MAX
116: };
117:
118: static const enum mdocargt args_Bk[] = {
119: MDOC_Words,
120: MDOC_ARG_MAX
121: };
122:
123: static const enum mdocargt args_Bl[] = {
124: MDOC_Bullet,
125: MDOC_Dash,
126: MDOC_Hyphen,
127: MDOC_Item,
128: MDOC_Enum,
129: MDOC_Tag,
130: MDOC_Diag,
131: MDOC_Hang,
132: MDOC_Ohang,
133: MDOC_Inset,
134: MDOC_Column,
135: MDOC_Width,
136: MDOC_Offset,
137: MDOC_Compact,
138: MDOC_Nested,
139: MDOC_ARG_MAX
140: };
141:
1.39 schwarze 142: static const struct mdocarg mdocargs[MDOC_MAX] = {
1.41 schwarze 143: { ARGSFL_DELIM, NULL }, /* Ap */
1.39 schwarze 144: { ARGSFL_NONE, NULL }, /* Dd */
145: { ARGSFL_NONE, NULL }, /* Dt */
146: { ARGSFL_NONE, NULL }, /* Os */
147: { ARGSFL_NONE, NULL }, /* Sh */
148: { ARGSFL_NONE, NULL }, /* Ss */
149: { ARGSFL_NONE, NULL }, /* Pp */
150: { ARGSFL_DELIM, NULL }, /* D1 */
151: { ARGSFL_DELIM, NULL }, /* Dl */
152: { ARGSFL_NONE, args_Bd }, /* Bd */
153: { ARGSFL_NONE, NULL }, /* Ed */
154: { ARGSFL_NONE, args_Bl }, /* Bl */
155: { ARGSFL_NONE, NULL }, /* El */
156: { ARGSFL_NONE, NULL }, /* It */
157: { ARGSFL_DELIM, NULL }, /* Ad */
158: { ARGSFL_DELIM, args_An }, /* An */
159: { ARGSFL_DELIM, NULL }, /* Ar */
1.43 schwarze 160: { ARGSFL_DELIM, NULL }, /* Cd */
1.39 schwarze 161: { ARGSFL_DELIM, NULL }, /* Cm */
162: { ARGSFL_DELIM, NULL }, /* Dv */
163: { ARGSFL_DELIM, NULL }, /* Er */
164: { ARGSFL_DELIM, NULL }, /* Ev */
165: { ARGSFL_NONE, args_Ex }, /* Ex */
166: { ARGSFL_DELIM, NULL }, /* Fa */
167: { ARGSFL_NONE, NULL }, /* Fd */
168: { ARGSFL_DELIM, NULL }, /* Fl */
169: { ARGSFL_DELIM, NULL }, /* Fn */
170: { ARGSFL_DELIM, NULL }, /* Ft */
171: { ARGSFL_DELIM, NULL }, /* Ic */
1.41 schwarze 172: { ARGSFL_DELIM, NULL }, /* In */
1.39 schwarze 173: { ARGSFL_DELIM, NULL }, /* Li */
174: { ARGSFL_NONE, NULL }, /* Nd */
175: { ARGSFL_DELIM, NULL }, /* Nm */
176: { ARGSFL_DELIM, NULL }, /* Op */
177: { ARGSFL_NONE, NULL }, /* Ot */
178: { ARGSFL_DELIM, NULL }, /* Pa */
179: { ARGSFL_NONE, args_Ex }, /* Rv */
180: { ARGSFL_DELIM, NULL }, /* St */
181: { ARGSFL_DELIM, NULL }, /* Va */
182: { ARGSFL_DELIM, NULL }, /* Vt */
183: { ARGSFL_DELIM, NULL }, /* Xr */
184: { ARGSFL_NONE, NULL }, /* %A */
185: { ARGSFL_NONE, NULL }, /* %B */
186: { ARGSFL_NONE, NULL }, /* %D */
187: { ARGSFL_NONE, NULL }, /* %I */
188: { ARGSFL_NONE, NULL }, /* %J */
189: { ARGSFL_NONE, NULL }, /* %N */
190: { ARGSFL_NONE, NULL }, /* %O */
191: { ARGSFL_NONE, NULL }, /* %P */
192: { ARGSFL_NONE, NULL }, /* %R */
193: { ARGSFL_NONE, NULL }, /* %T */
194: { ARGSFL_NONE, NULL }, /* %V */
195: { ARGSFL_DELIM, NULL }, /* Ac */
196: { ARGSFL_NONE, NULL }, /* Ao */
197: { ARGSFL_DELIM, NULL }, /* Aq */
198: { ARGSFL_DELIM, NULL }, /* At */
199: { ARGSFL_DELIM, NULL }, /* Bc */
200: { ARGSFL_NONE, args_Bf }, /* Bf */
201: { ARGSFL_NONE, NULL }, /* Bo */
202: { ARGSFL_DELIM, NULL }, /* Bq */
203: { ARGSFL_DELIM, NULL }, /* Bsx */
204: { ARGSFL_DELIM, NULL }, /* Bx */
205: { ARGSFL_NONE, NULL }, /* Db */
206: { ARGSFL_DELIM, NULL }, /* Dc */
207: { ARGSFL_NONE, NULL }, /* Do */
208: { ARGSFL_DELIM, NULL }, /* Dq */
209: { ARGSFL_DELIM, NULL }, /* Ec */
210: { ARGSFL_NONE, NULL }, /* Ef */
211: { ARGSFL_DELIM, NULL }, /* Em */
212: { ARGSFL_NONE, NULL }, /* Eo */
213: { ARGSFL_DELIM, NULL }, /* Fx */
214: { ARGSFL_DELIM, NULL }, /* Ms */
215: { ARGSFL_DELIM, NULL }, /* No */
216: { ARGSFL_DELIM, NULL }, /* Ns */
217: { ARGSFL_DELIM, NULL }, /* Nx */
218: { ARGSFL_DELIM, NULL }, /* Ox */
219: { ARGSFL_DELIM, NULL }, /* Pc */
220: { ARGSFL_DELIM, NULL }, /* Pf */
221: { ARGSFL_NONE, NULL }, /* Po */
222: { ARGSFL_DELIM, NULL }, /* Pq */
223: { ARGSFL_DELIM, NULL }, /* Qc */
224: { ARGSFL_DELIM, NULL }, /* Ql */
225: { ARGSFL_NONE, NULL }, /* Qo */
226: { ARGSFL_DELIM, NULL }, /* Qq */
227: { ARGSFL_NONE, NULL }, /* Re */
228: { ARGSFL_NONE, NULL }, /* Rs */
229: { ARGSFL_DELIM, NULL }, /* Sc */
230: { ARGSFL_NONE, NULL }, /* So */
231: { ARGSFL_DELIM, NULL }, /* Sq */
232: { ARGSFL_NONE, NULL }, /* Sm */
233: { ARGSFL_DELIM, NULL }, /* Sx */
234: { ARGSFL_DELIM, NULL }, /* Sy */
235: { ARGSFL_DELIM, NULL }, /* Tn */
236: { ARGSFL_DELIM, NULL }, /* Ux */
237: { ARGSFL_DELIM, NULL }, /* Xc */
238: { ARGSFL_NONE, NULL }, /* Xo */
239: { ARGSFL_NONE, NULL }, /* Fo */
1.43 schwarze 240: { ARGSFL_DELIM, NULL }, /* Fc */
1.39 schwarze 241: { ARGSFL_NONE, NULL }, /* Oo */
242: { ARGSFL_DELIM, NULL }, /* Oc */
243: { ARGSFL_NONE, args_Bk }, /* Bk */
244: { ARGSFL_NONE, NULL }, /* Ek */
245: { ARGSFL_NONE, NULL }, /* Bt */
246: { ARGSFL_NONE, NULL }, /* Hf */
247: { ARGSFL_NONE, NULL }, /* Fr */
248: { ARGSFL_NONE, NULL }, /* Ud */
1.43 schwarze 249: { ARGSFL_DELIM, NULL }, /* Lb */
1.39 schwarze 250: { ARGSFL_NONE, NULL }, /* Lp */
251: { ARGSFL_DELIM, NULL }, /* Lk */
252: { ARGSFL_DELIM, NULL }, /* Mt */
253: { ARGSFL_DELIM, NULL }, /* Brq */
254: { ARGSFL_NONE, NULL }, /* Bro */
255: { ARGSFL_DELIM, NULL }, /* Brc */
256: { ARGSFL_NONE, NULL }, /* %C */
257: { ARGSFL_NONE, NULL }, /* Es */
258: { ARGSFL_NONE, NULL }, /* En */
1.45 schwarze 259: { ARGSFL_DELIM, NULL }, /* Dx */
1.39 schwarze 260: { ARGSFL_NONE, NULL }, /* %Q */
261: { ARGSFL_NONE, NULL }, /* br */
262: { ARGSFL_NONE, NULL }, /* sp */
263: { ARGSFL_NONE, NULL }, /* %U */
264: { ARGSFL_NONE, NULL }, /* Ta */
265: };
266:
267:
1.1 kristaps 268: /*
269: * Parse an argument from line text. This comes in the form of -key
270: * [value0...], which may either have a single mandatory value, at least
271: * one mandatory value, an optional single value, or no value.
272: */
1.24 schwarze 273: enum margverr
1.42 schwarze 274: mdoc_argv(struct mdoc *mdoc, int line, enum mdoct tok,
1.1 kristaps 275: struct mdoc_arg **v, int *pos, char *buf)
276: {
277: char *p, sv;
278: struct mdoc_argv tmp;
279: struct mdoc_arg *arg;
1.39 schwarze 280: const enum mdocargt *ap;
1.1 kristaps 281:
1.30 schwarze 282: if ('\0' == buf[*pos])
1.1 kristaps 283: return(ARGV_EOLN);
1.39 schwarze 284: else if (NULL == (ap = mdocargs[tok].argvs))
285: return(ARGV_WORD);
286: else if ('-' != buf[*pos])
287: return(ARGV_WORD);
1.1 kristaps 288:
1.39 schwarze 289: /* Seek to the first unescaped space. */
1.1 kristaps 290:
291: p = &buf[++(*pos)];
292:
293: assert(*pos > 0);
294:
1.39 schwarze 295: for ( ; buf[*pos] ; (*pos)++)
296: if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
297: break;
1.1 kristaps 298:
1.39 schwarze 299: /*
300: * We want to nil-terminate the word to look it up (it's easier
301: * that way). But we may not have a flag, in which case we need
302: * to restore the line as-is. So keep around the stray byte,
303: * which we'll reset upon exiting (if necessary).
304: */
1.1 kristaps 305:
1.39 schwarze 306: if ('\0' != (sv = buf[*pos]))
1.30 schwarze 307: buf[(*pos)++] = '\0';
1.39 schwarze 308:
309: /*
310: * Now look up the word as a flag. Use temporary storage that
311: * we'll copy into the node's flags, if necessary.
312: */
1.1 kristaps 313:
1.36 schwarze 314: memset(&tmp, 0, sizeof(struct mdoc_argv));
1.39 schwarze 315:
1.1 kristaps 316: tmp.line = line;
317: tmp.pos = *pos;
1.39 schwarze 318: tmp.arg = MDOC_ARG_MAX;
1.1 kristaps 319:
1.39 schwarze 320: while (MDOC_ARG_MAX != (tmp.arg = *ap++))
321: if (0 == strcmp(p, mdoc_argnames[tmp.arg]))
322: break;
1.1 kristaps 323:
1.39 schwarze 324: if (MDOC_ARG_MAX == tmp.arg) {
325: /*
326: * The flag was not found.
327: * Restore saved zeroed byte and return as a word.
328: */
1.1 kristaps 329: if (sv)
330: buf[*pos - 1] = sv;
331: return(ARGV_WORD);
332: }
333:
1.39 schwarze 334: /* Read to the next word (the argument). */
335:
1.1 kristaps 336: while (buf[*pos] && ' ' == buf[*pos])
337: (*pos)++;
338:
1.39 schwarze 339: switch (argvflags[tmp.arg]) {
340: case (ARGV_SINGLE):
1.42 schwarze 341: if ( ! argv_single(mdoc, line, &tmp, pos, buf))
1.39 schwarze 342: return(ARGV_ERROR);
343: break;
344: case (ARGV_MULTI):
1.42 schwarze 345: if ( ! argv_multi(mdoc, line, &tmp, pos, buf))
1.39 schwarze 346: return(ARGV_ERROR);
347: break;
348: case (ARGV_NONE):
349: break;
350: }
1.1 kristaps 351:
1.19 schwarze 352: if (NULL == (arg = *v))
353: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 354:
355: arg->argc++;
1.19 schwarze 356: arg->argv = mandoc_realloc
357: (arg->argv, arg->argc * sizeof(struct mdoc_argv));
1.1 kristaps 358:
1.36 schwarze 359: memcpy(&arg->argv[(int)arg->argc - 1],
1.1 kristaps 360: &tmp, sizeof(struct mdoc_argv));
361:
362: return(ARGV_ARG);
363: }
364:
365: void
366: mdoc_argv_free(struct mdoc_arg *p)
367: {
1.23 schwarze 368: int i;
1.1 kristaps 369:
370: if (NULL == p)
371: return;
372:
373: if (p->refcnt) {
374: --(p->refcnt);
375: if (p->refcnt)
376: return;
377: }
378: assert(p->argc);
379:
1.24 schwarze 380: for (i = (int)p->argc - 1; i >= 0; i--)
1.37 schwarze 381: argn_free(p, i);
1.23 schwarze 382:
383: free(p->argv);
384: free(p);
385: }
386:
1.37 schwarze 387: static void
388: argn_free(struct mdoc_arg *p, int iarg)
1.23 schwarze 389: {
1.33 schwarze 390: struct mdoc_argv *arg;
1.23 schwarze 391: int j;
1.1 kristaps 392:
1.33 schwarze 393: arg = &p->argv[iarg];
394:
1.23 schwarze 395: if (arg->sz && arg->value) {
1.24 schwarze 396: for (j = (int)arg->sz - 1; j >= 0; j--)
1.23 schwarze 397: free(arg->value[j]);
398: free(arg->value);
1.1 kristaps 399: }
400:
1.23 schwarze 401: for (--p->argc; iarg < (int)p->argc; iarg++)
402: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 403: }
404:
1.24 schwarze 405: enum margserr
1.42 schwarze 406: mdoc_zargs(struct mdoc *mdoc, int line, int *pos, char *buf, char **v)
1.12 schwarze 407: {
408:
1.42 schwarze 409: return(args(mdoc, line, pos, buf, ARGSFL_NONE, v));
1.12 schwarze 410: }
411:
1.24 schwarze 412: enum margserr
1.42 schwarze 413: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.22 schwarze 414: char *buf, enum mdoct tok, char **v)
1.1 kristaps 415: {
1.38 schwarze 416: enum argsflag fl;
1.1 kristaps 417: struct mdoc_node *n;
418:
1.39 schwarze 419: fl = mdocargs[tok].flags;
1.1 kristaps 420:
1.12 schwarze 421: if (MDOC_It != tok)
1.42 schwarze 422: return(args(mdoc, line, pos, buf, fl, v));
1.12 schwarze 423:
1.30 schwarze 424: /*
425: * We know that we're in an `It', so it's reasonable to expect
426: * us to be sitting in a `Bl'. Someday this may not be the case
427: * (if we allow random `It's sitting out there), so provide a
428: * safe fall-back into the default behaviour.
1.1 kristaps 429: */
430:
1.42 schwarze 431: for (n = mdoc->last; n; n = n->parent)
1.30 schwarze 432: if (MDOC_Bl == n->tok)
1.38 schwarze 433: if (LIST_column == n->norm->Bl.type) {
434: fl = ARGSFL_TABSEP;
435: break;
436: }
1.1 kristaps 437:
1.42 schwarze 438: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 439: }
440:
1.24 schwarze 441: static enum margserr
1.42 schwarze 442: args(struct mdoc *mdoc, int line, int *pos,
1.38 schwarze 443: char *buf, enum argsflag fl, char **v)
1.1 kristaps 444: {
1.30 schwarze 445: char *p, *pp;
1.44 schwarze 446: int pairs;
1.30 schwarze 447: enum margserr rc;
1.1 kristaps 448:
1.28 schwarze 449: if ('\0' == buf[*pos]) {
1.42 schwarze 450: if (MDOC_PPHRASE & mdoc->flags)
1.28 schwarze 451: return(ARGS_EOLN);
452: /*
453: * If we're not in a partial phrase and the flag for
454: * being a phrase literal is still set, the punctuation
455: * is unterminated.
456: */
1.42 schwarze 457: if (MDOC_PHRASELIT & mdoc->flags)
458: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
1.28 schwarze 459:
1.42 schwarze 460: mdoc->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 461: return(ARGS_EOLN);
1.28 schwarze 462: }
1.1 kristaps 463:
1.36 schwarze 464: *v = &buf[*pos];
1.1 kristaps 465:
1.38 schwarze 466: if (ARGSFL_DELIM == fl)
467: if (args_checkpunct(buf, *pos))
1.36 schwarze 468: return(ARGS_PUNCT);
1.1 kristaps 469:
1.12 schwarze 470: /*
471: * First handle TABSEP items, restricted to `Bl -column'. This
472: * ignores conventional token parsing and instead uses tabs or
473: * `Ta' macros to separate phrases. Phrases are parsed again
474: * for arguments at a later phase.
475: */
476:
1.38 schwarze 477: if (ARGSFL_TABSEP == fl) {
1.12 schwarze 478: /* Scan ahead to tab (can't be escaped). */
479: p = strchr(*v, '\t');
1.28 schwarze 480: pp = NULL;
1.12 schwarze 481:
482: /* Scan ahead to unescaped `Ta'. */
1.42 schwarze 483: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.28 schwarze 484: for (pp = *v; ; pp++) {
485: if (NULL == (pp = strstr(pp, "Ta")))
486: break;
487: if (pp > *v && ' ' != *(pp - 1))
488: continue;
1.30 schwarze 489: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.28 schwarze 490: break;
491: }
1.1 kristaps 492:
1.24 schwarze 493: /* By default, assume a phrase. */
494: rc = ARGS_PHRASE;
495:
1.1 kristaps 496: /*
1.12 schwarze 497: * Adjust new-buffer position to be beyond delimiter
498: * mark (e.g., Ta -> end + 2).
1.1 kristaps 499: */
1.12 schwarze 500: if (p && pp) {
501: *pos += pp < p ? 2 : 1;
1.24 schwarze 502: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.12 schwarze 503: p = pp < p ? pp : p;
504: } else if (p && ! pp) {
1.24 schwarze 505: rc = ARGS_PPHRASE;
1.12 schwarze 506: *pos += 1;
507: } else if (pp && ! p) {
508: p = pp;
509: *pos += 2;
1.25 schwarze 510: } else {
511: rc = ARGS_PEND;
1.12 schwarze 512: p = strchr(*v, 0);
1.25 schwarze 513: }
1.1 kristaps 514:
1.12 schwarze 515: /* Whitespace check for eoln case... */
1.38 schwarze 516: if ('\0' == *p && ' ' == *(p - 1))
1.42 schwarze 517: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
1.12 schwarze 518:
519: *pos += (int)(p - *v);
520:
521: /* Strip delimiter's preceding whitespace. */
522: pp = p - 1;
523: while (pp > *v && ' ' == *pp) {
524: if (pp > *v && '\\' == *(pp - 1))
525: break;
526: pp--;
527: }
528: *(pp + 1) = 0;
1.1 kristaps 529:
1.12 schwarze 530: /* Strip delimiter's proceeding whitespace. */
531: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
532: /* Skip ahead. */ ;
1.1 kristaps 533:
1.24 schwarze 534: return(rc);
1.46 ! schwarze 535: }
1.1 kristaps 536:
1.46 ! schwarze 537: /*
1.12 schwarze 538: * Process a quoted literal. A quote begins with a double-quote
539: * and ends with a double-quote NOT preceded by a double-quote.
1.46 ! schwarze 540: * NUL-terminate the literal in place.
1.44 schwarze 541: * Collapse pairs of quotes inside quoted literals.
1.12 schwarze 542: * Whitespace is NOT involved in literal termination.
543: */
1.1 kristaps 544:
1.42 schwarze 545: if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
546: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.28 schwarze 547: *v = &buf[++(*pos)];
548:
1.42 schwarze 549: if (MDOC_PPHRASE & mdoc->flags)
550: mdoc->flags |= MDOC_PHRASELIT;
1.1 kristaps 551:
1.44 schwarze 552: pairs = 0;
1.12 schwarze 553: for ( ; buf[*pos]; (*pos)++) {
1.44 schwarze 554: /* Move following text left after quoted quotes. */
555: if (pairs)
556: buf[*pos - pairs] = buf[*pos];
1.12 schwarze 557: if ('\"' != buf[*pos])
558: continue;
1.44 schwarze 559: /* Unquoted quotes end quoted args. */
1.12 schwarze 560: if ('\"' != buf[*pos + 1])
561: break;
1.44 schwarze 562: /* Quoted quotes collapse. */
563: pairs++;
1.12 schwarze 564: (*pos)++;
565: }
1.44 schwarze 566: if (pairs)
567: buf[*pos - pairs] = '\0';
1.1 kristaps 568:
1.28 schwarze 569: if ('\0' == buf[*pos]) {
1.42 schwarze 570: if (MDOC_PPHRASE & mdoc->flags)
1.13 schwarze 571: return(ARGS_QWORD);
1.42 schwarze 572: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
1.12 schwarze 573: return(ARGS_QWORD);
574: }
1.1 kristaps 575:
1.42 schwarze 576: mdoc->flags &= ~MDOC_PHRASELIT;
1.28 schwarze 577: buf[(*pos)++] = '\0';
1.1 kristaps 578:
1.28 schwarze 579: if ('\0' == buf[*pos])
1.12 schwarze 580: return(ARGS_QWORD);
1.1 kristaps 581:
1.12 schwarze 582: while (' ' == buf[*pos])
583: (*pos)++;
1.1 kristaps 584:
1.38 schwarze 585: if ('\0' == buf[*pos])
1.42 schwarze 586: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 587:
1.12 schwarze 588: return(ARGS_QWORD);
1.1 kristaps 589: }
590:
1.38 schwarze 591: p = &buf[*pos];
1.42 schwarze 592: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 593:
1.12 schwarze 594: return(ARGS_WORD);
1.1 kristaps 595: }
596:
1.36 schwarze 597: /*
598: * Check if the string consists only of space-separated closing
599: * delimiters. This is a bit of a dance: the first must be a close
600: * delimiter, but it may be followed by middle delimiters. Arbitrary
601: * whitespace may separate these tokens.
602: */
603: static int
1.38 schwarze 604: args_checkpunct(const char *buf, int i)
1.36 schwarze 605: {
1.38 schwarze 606: int j;
607: char dbuf[DELIMSZ];
1.36 schwarze 608: enum mdelim d;
609:
610: /* First token must be a close-delimiter. */
611:
1.38 schwarze 612: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
613: dbuf[j] = buf[i];
1.1 kristaps 614:
1.36 schwarze 615: if (DELIMSZ == j)
616: return(0);
617:
1.38 schwarze 618: dbuf[j] = '\0';
619: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.36 schwarze 620: return(0);
621:
1.38 schwarze 622: while (' ' == buf[i])
1.36 schwarze 623: i++;
624:
625: /* Remaining must NOT be open/none. */
626:
1.38 schwarze 627: while (buf[i]) {
1.36 schwarze 628: j = 0;
1.38 schwarze 629: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
630: dbuf[j++] = buf[i++];
1.36 schwarze 631:
632: if (DELIMSZ == j)
633: return(0);
634:
1.38 schwarze 635: dbuf[j] = '\0';
636: d = mdoc_isdelim(dbuf);
1.36 schwarze 637: if (DELIM_NONE == d || DELIM_OPEN == d)
638: return(0);
639:
1.38 schwarze 640: while (' ' == buf[i])
1.36 schwarze 641: i++;
642: }
643:
1.38 schwarze 644: return('\0' == buf[i]);
1.36 schwarze 645: }
646:
1.1 kristaps 647: static int
1.42 schwarze 648: argv_multi(struct mdoc *mdoc, int line,
1.1 kristaps 649: struct mdoc_argv *v, int *pos, char *buf)
650: {
1.24 schwarze 651: enum margserr ac;
1.1 kristaps 652: char *p;
653:
654: for (v->sz = 0; ; v->sz++) {
655: if ('-' == buf[*pos])
656: break;
1.42 schwarze 657: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.24 schwarze 658: if (ARGS_ERROR == ac)
1.1 kristaps 659: return(0);
1.24 schwarze 660: else if (ARGS_EOLN == ac)
1.1 kristaps 661: break;
662:
1.19 schwarze 663: if (0 == v->sz % MULTI_STEP)
664: v->value = mandoc_realloc(v->value,
1.1 kristaps 665: (v->sz + MULTI_STEP) * sizeof(char *));
1.19 schwarze 666:
667: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 668: }
669:
1.4 schwarze 670: return(1);
1.1 kristaps 671: }
672:
673: static int
1.42 schwarze 674: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 675: struct mdoc_argv *v, int *pos, char *buf)
676: {
1.24 schwarze 677: enum margserr ac;
1.1 kristaps 678: char *p;
679:
1.42 schwarze 680: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.24 schwarze 681: if (ARGS_ERROR == ac)
1.1 kristaps 682: return(0);
1.24 schwarze 683: if (ARGS_EOLN == ac)
1.1 kristaps 684: return(1);
685:
686: v->sz = 1;
1.19 schwarze 687: v->value = mandoc_malloc(sizeof(char *));
688: v->value[0] = mandoc_strdup(p);
1.1 kristaps 689:
690: return(1);
691: }