Annotation of src/usr.bin/mandoc/mdoc_argv.c, Revision 1.50
1.50 ! schwarze 1: /* $Id: mdoc_argv.c,v 1.49 2014/04/20 16:44:44 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"
1.47 schwarze 27: #include "mandoc_aux.h"
1.1 kristaps 28: #include "libmdoc.h"
1.19 schwarze 29: #include "libmandoc.h"
1.1 kristaps 30:
1.36 schwarze 31: #define MULTI_STEP 5 /* pre-allocate argument values */
1.49 schwarze 32: #define DELIMSZ 6 /* max possible size of a delimiter */
1.38 schwarze 33:
34: enum argsflag {
35: ARGSFL_NONE = 0,
36: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
37: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
38: };
39:
40: enum argvflag {
41: ARGV_NONE, /* no args to flag (e.g., -split) */
42: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
1.40 schwarze 43: ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
1.38 schwarze 44: };
1.1 kristaps 45:
1.39 schwarze 46: struct mdocarg {
47: enum argsflag flags;
48: const enum mdocargt *argvs;
49: };
50:
51: static void argn_free(struct mdoc_arg *, int);
1.49 schwarze 52: static enum margserr args(struct mdoc *, int, int *,
1.38 schwarze 53: char *, enum argsflag, char **);
54: static int args_checkpunct(const char *, int);
1.49 schwarze 55: static int argv_multi(struct mdoc *, int,
1.1 kristaps 56: struct mdoc_argv *, int *, char *);
1.49 schwarze 57: static int argv_single(struct mdoc *, int,
1.1 kristaps 58: struct mdoc_argv *, int *, char *);
59:
1.36 schwarze 60: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 61: ARGV_NONE, /* MDOC_Split */
62: ARGV_NONE, /* MDOC_Nosplit */
63: ARGV_NONE, /* MDOC_Ragged */
64: ARGV_NONE, /* MDOC_Unfilled */
65: ARGV_NONE, /* MDOC_Literal */
1.17 schwarze 66: ARGV_SINGLE, /* MDOC_File */
1.40 schwarze 67: ARGV_SINGLE, /* MDOC_Offset */
1.1 kristaps 68: ARGV_NONE, /* MDOC_Bullet */
69: ARGV_NONE, /* MDOC_Dash */
70: ARGV_NONE, /* MDOC_Hyphen */
71: ARGV_NONE, /* MDOC_Item */
72: ARGV_NONE, /* MDOC_Enum */
73: ARGV_NONE, /* MDOC_Tag */
74: ARGV_NONE, /* MDOC_Diag */
75: ARGV_NONE, /* MDOC_Hang */
76: ARGV_NONE, /* MDOC_Ohang */
77: ARGV_NONE, /* MDOC_Inset */
78: ARGV_MULTI, /* MDOC_Column */
79: ARGV_SINGLE, /* MDOC_Width */
80: ARGV_NONE, /* MDOC_Compact */
1.14 schwarze 81: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 82: ARGV_NONE, /* MDOC_Filled */
83: ARGV_NONE, /* MDOC_Words */
84: ARGV_NONE, /* MDOC_Emphasis */
85: ARGV_NONE, /* MDOC_Symbolic */
86: ARGV_NONE /* MDOC_Symbolic */
87: };
88:
1.37 schwarze 89: static const enum mdocargt args_Ex[] = {
90: MDOC_Std,
91: MDOC_ARG_MAX
92: };
93:
94: static const enum mdocargt args_An[] = {
95: MDOC_Split,
96: MDOC_Nosplit,
97: MDOC_ARG_MAX
98: };
99:
100: static const enum mdocargt args_Bd[] = {
101: MDOC_Ragged,
102: MDOC_Unfilled,
103: MDOC_Filled,
104: MDOC_Literal,
105: MDOC_File,
106: MDOC_Offset,
107: MDOC_Compact,
108: MDOC_Centred,
109: MDOC_ARG_MAX
110: };
111:
112: static const enum mdocargt args_Bf[] = {
113: MDOC_Emphasis,
114: MDOC_Literal,
115: MDOC_Symbolic,
116: MDOC_ARG_MAX
117: };
118:
119: static const enum mdocargt args_Bk[] = {
120: MDOC_Words,
121: MDOC_ARG_MAX
122: };
123:
124: static const enum mdocargt args_Bl[] = {
125: MDOC_Bullet,
126: MDOC_Dash,
127: MDOC_Hyphen,
128: MDOC_Item,
129: MDOC_Enum,
130: MDOC_Tag,
131: MDOC_Diag,
132: MDOC_Hang,
133: MDOC_Ohang,
134: MDOC_Inset,
135: MDOC_Column,
136: MDOC_Width,
137: MDOC_Offset,
138: MDOC_Compact,
139: MDOC_Nested,
140: MDOC_ARG_MAX
141: };
142:
1.39 schwarze 143: static const struct mdocarg mdocargs[MDOC_MAX] = {
1.41 schwarze 144: { ARGSFL_DELIM, NULL }, /* Ap */
1.39 schwarze 145: { ARGSFL_NONE, NULL }, /* Dd */
146: { ARGSFL_NONE, NULL }, /* Dt */
147: { ARGSFL_NONE, NULL }, /* Os */
148: { ARGSFL_NONE, NULL }, /* Sh */
1.49 schwarze 149: { ARGSFL_NONE, NULL }, /* Ss */
150: { ARGSFL_NONE, NULL }, /* Pp */
1.39 schwarze 151: { ARGSFL_DELIM, NULL }, /* D1 */
152: { ARGSFL_DELIM, NULL }, /* Dl */
153: { ARGSFL_NONE, args_Bd }, /* Bd */
154: { ARGSFL_NONE, NULL }, /* Ed */
155: { ARGSFL_NONE, args_Bl }, /* Bl */
156: { ARGSFL_NONE, NULL }, /* El */
157: { ARGSFL_NONE, NULL }, /* It */
1.49 schwarze 158: { ARGSFL_DELIM, NULL }, /* Ad */
1.39 schwarze 159: { ARGSFL_DELIM, args_An }, /* An */
160: { ARGSFL_DELIM, NULL }, /* Ar */
1.43 schwarze 161: { ARGSFL_DELIM, NULL }, /* Cd */
1.39 schwarze 162: { ARGSFL_DELIM, NULL }, /* Cm */
1.49 schwarze 163: { ARGSFL_DELIM, NULL }, /* Dv */
164: { ARGSFL_DELIM, NULL }, /* Er */
165: { ARGSFL_DELIM, NULL }, /* Ev */
1.39 schwarze 166: { ARGSFL_NONE, args_Ex }, /* Ex */
1.49 schwarze 167: { ARGSFL_DELIM, NULL }, /* Fa */
168: { ARGSFL_NONE, NULL }, /* Fd */
1.39 schwarze 169: { ARGSFL_DELIM, NULL }, /* Fl */
1.49 schwarze 170: { ARGSFL_DELIM, NULL }, /* Fn */
171: { ARGSFL_DELIM, NULL }, /* Ft */
172: { ARGSFL_DELIM, NULL }, /* Ic */
173: { ARGSFL_DELIM, NULL }, /* In */
1.39 schwarze 174: { ARGSFL_DELIM, NULL }, /* Li */
1.49 schwarze 175: { ARGSFL_NONE, NULL }, /* Nd */
176: { ARGSFL_DELIM, NULL }, /* Nm */
1.39 schwarze 177: { ARGSFL_DELIM, NULL }, /* Op */
178: { ARGSFL_NONE, NULL }, /* Ot */
179: { ARGSFL_DELIM, NULL }, /* Pa */
180: { ARGSFL_NONE, args_Ex }, /* Rv */
1.49 schwarze 181: { ARGSFL_DELIM, NULL }, /* St */
1.39 schwarze 182: { ARGSFL_DELIM, NULL }, /* Va */
1.49 schwarze 183: { ARGSFL_DELIM, NULL }, /* Vt */
1.39 schwarze 184: { ARGSFL_DELIM, NULL }, /* Xr */
185: { ARGSFL_NONE, NULL }, /* %A */
186: { ARGSFL_NONE, NULL }, /* %B */
187: { ARGSFL_NONE, NULL }, /* %D */
188: { ARGSFL_NONE, NULL }, /* %I */
189: { ARGSFL_NONE, NULL }, /* %J */
190: { ARGSFL_NONE, NULL }, /* %N */
191: { ARGSFL_NONE, NULL }, /* %O */
192: { ARGSFL_NONE, NULL }, /* %P */
193: { ARGSFL_NONE, NULL }, /* %R */
194: { ARGSFL_NONE, NULL }, /* %T */
195: { ARGSFL_NONE, NULL }, /* %V */
196: { ARGSFL_DELIM, NULL }, /* Ac */
197: { ARGSFL_NONE, NULL }, /* Ao */
198: { ARGSFL_DELIM, NULL }, /* Aq */
199: { ARGSFL_DELIM, NULL }, /* At */
200: { ARGSFL_DELIM, NULL }, /* Bc */
1.49 schwarze 201: { ARGSFL_NONE, args_Bf }, /* Bf */
1.39 schwarze 202: { ARGSFL_NONE, NULL }, /* Bo */
203: { ARGSFL_DELIM, NULL }, /* Bq */
204: { ARGSFL_DELIM, NULL }, /* Bsx */
205: { ARGSFL_DELIM, NULL }, /* Bx */
206: { ARGSFL_NONE, NULL }, /* Db */
207: { ARGSFL_DELIM, NULL }, /* Dc */
208: { ARGSFL_NONE, NULL }, /* Do */
209: { ARGSFL_DELIM, NULL }, /* Dq */
210: { ARGSFL_DELIM, NULL }, /* Ec */
211: { ARGSFL_NONE, NULL }, /* Ef */
1.49 schwarze 212: { ARGSFL_DELIM, NULL }, /* Em */
1.39 schwarze 213: { ARGSFL_NONE, NULL }, /* Eo */
214: { ARGSFL_DELIM, NULL }, /* Fx */
215: { ARGSFL_DELIM, NULL }, /* Ms */
216: { ARGSFL_DELIM, NULL }, /* No */
217: { ARGSFL_DELIM, NULL }, /* Ns */
218: { ARGSFL_DELIM, NULL }, /* Nx */
219: { ARGSFL_DELIM, NULL }, /* Ox */
220: { ARGSFL_DELIM, NULL }, /* Pc */
221: { ARGSFL_DELIM, NULL }, /* Pf */
222: { ARGSFL_NONE, NULL }, /* Po */
223: { ARGSFL_DELIM, NULL }, /* Pq */
224: { ARGSFL_DELIM, NULL }, /* Qc */
225: { ARGSFL_DELIM, NULL }, /* Ql */
226: { ARGSFL_NONE, NULL }, /* Qo */
227: { ARGSFL_DELIM, NULL }, /* Qq */
228: { ARGSFL_NONE, NULL }, /* Re */
229: { ARGSFL_NONE, NULL }, /* Rs */
230: { ARGSFL_DELIM, NULL }, /* Sc */
231: { ARGSFL_NONE, NULL }, /* So */
232: { ARGSFL_DELIM, NULL }, /* Sq */
233: { ARGSFL_NONE, NULL }, /* Sm */
234: { ARGSFL_DELIM, NULL }, /* Sx */
235: { ARGSFL_DELIM, NULL }, /* Sy */
236: { ARGSFL_DELIM, NULL }, /* Tn */
237: { ARGSFL_DELIM, NULL }, /* Ux */
238: { ARGSFL_DELIM, NULL }, /* Xc */
239: { ARGSFL_NONE, NULL }, /* Xo */
1.49 schwarze 240: { ARGSFL_NONE, NULL }, /* Fo */
241: { ARGSFL_DELIM, NULL }, /* Fc */
1.39 schwarze 242: { ARGSFL_NONE, NULL }, /* Oo */
243: { ARGSFL_DELIM, NULL }, /* Oc */
244: { ARGSFL_NONE, args_Bk }, /* Bk */
245: { ARGSFL_NONE, NULL }, /* Ek */
246: { ARGSFL_NONE, NULL }, /* Bt */
247: { ARGSFL_NONE, NULL }, /* Hf */
248: { ARGSFL_NONE, NULL }, /* Fr */
249: { ARGSFL_NONE, NULL }, /* Ud */
1.43 schwarze 250: { ARGSFL_DELIM, NULL }, /* Lb */
1.39 schwarze 251: { ARGSFL_NONE, NULL }, /* Lp */
252: { ARGSFL_DELIM, NULL }, /* Lk */
253: { ARGSFL_DELIM, NULL }, /* Mt */
254: { ARGSFL_DELIM, NULL }, /* Brq */
255: { ARGSFL_NONE, NULL }, /* Bro */
256: { ARGSFL_DELIM, NULL }, /* Brc */
257: { ARGSFL_NONE, NULL }, /* %C */
258: { ARGSFL_NONE, NULL }, /* Es */
259: { ARGSFL_NONE, NULL }, /* En */
1.45 schwarze 260: { ARGSFL_DELIM, NULL }, /* Dx */
1.39 schwarze 261: { ARGSFL_NONE, NULL }, /* %Q */
262: { ARGSFL_NONE, NULL }, /* br */
263: { ARGSFL_NONE, NULL }, /* sp */
264: { ARGSFL_NONE, NULL }, /* %U */
265: { ARGSFL_NONE, NULL }, /* Ta */
1.48 schwarze 266: { ARGSFL_NONE, NULL }, /* ll */
1.39 schwarze 267: };
268:
269:
1.1 kristaps 270: /*
271: * Parse an argument from line text. This comes in the form of -key
272: * [value0...], which may either have a single mandatory value, at least
273: * one mandatory value, an optional single value, or no value.
274: */
1.24 schwarze 275: enum margverr
1.42 schwarze 276: mdoc_argv(struct mdoc *mdoc, int line, enum mdoct tok,
1.1 kristaps 277: struct mdoc_arg **v, int *pos, char *buf)
278: {
279: char *p, sv;
280: struct mdoc_argv tmp;
281: struct mdoc_arg *arg;
1.39 schwarze 282: const enum mdocargt *ap;
1.1 kristaps 283:
1.30 schwarze 284: if ('\0' == buf[*pos])
1.1 kristaps 285: return(ARGV_EOLN);
1.39 schwarze 286: else if (NULL == (ap = mdocargs[tok].argvs))
287: return(ARGV_WORD);
288: else if ('-' != buf[*pos])
289: return(ARGV_WORD);
1.1 kristaps 290:
1.39 schwarze 291: /* Seek to the first unescaped space. */
1.1 kristaps 292:
293: p = &buf[++(*pos)];
294:
295: assert(*pos > 0);
296:
1.39 schwarze 297: for ( ; buf[*pos] ; (*pos)++)
298: if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
299: break;
1.1 kristaps 300:
1.49 schwarze 301: /*
1.39 schwarze 302: * We want to nil-terminate the word to look it up (it's easier
303: * that way). But we may not have a flag, in which case we need
304: * to restore the line as-is. So keep around the stray byte,
305: * which we'll reset upon exiting (if necessary).
306: */
1.1 kristaps 307:
1.49 schwarze 308: if ('\0' != (sv = buf[*pos]))
1.30 schwarze 309: buf[(*pos)++] = '\0';
1.39 schwarze 310:
311: /*
312: * Now look up the word as a flag. Use temporary storage that
313: * we'll copy into the node's flags, if necessary.
314: */
1.1 kristaps 315:
1.36 schwarze 316: memset(&tmp, 0, sizeof(struct mdoc_argv));
1.39 schwarze 317:
1.1 kristaps 318: tmp.line = line;
319: tmp.pos = *pos;
1.39 schwarze 320: tmp.arg = MDOC_ARG_MAX;
1.1 kristaps 321:
1.39 schwarze 322: while (MDOC_ARG_MAX != (tmp.arg = *ap++))
323: if (0 == strcmp(p, mdoc_argnames[tmp.arg]))
324: break;
1.1 kristaps 325:
1.39 schwarze 326: if (MDOC_ARG_MAX == tmp.arg) {
1.49 schwarze 327: /*
1.39 schwarze 328: * The flag was not found.
329: * Restore saved zeroed byte and return as a word.
330: */
1.1 kristaps 331: if (sv)
332: buf[*pos - 1] = sv;
333: return(ARGV_WORD);
334: }
335:
1.39 schwarze 336: /* Read to the next word (the argument). */
337:
1.1 kristaps 338: while (buf[*pos] && ' ' == buf[*pos])
339: (*pos)++;
340:
1.39 schwarze 341: switch (argvflags[tmp.arg]) {
1.49 schwarze 342: case ARGV_SINGLE:
1.42 schwarze 343: if ( ! argv_single(mdoc, line, &tmp, pos, buf))
1.39 schwarze 344: return(ARGV_ERROR);
345: break;
1.49 schwarze 346: case ARGV_MULTI:
1.42 schwarze 347: if ( ! argv_multi(mdoc, line, &tmp, pos, buf))
1.39 schwarze 348: return(ARGV_ERROR);
349: break;
1.49 schwarze 350: case ARGV_NONE:
1.39 schwarze 351: break;
352: }
1.1 kristaps 353:
1.19 schwarze 354: if (NULL == (arg = *v))
355: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 356:
357: arg->argc++;
1.50 ! schwarze 358: arg->argv = mandoc_reallocarray(arg->argv,
! 359: arg->argc, sizeof(struct mdoc_argv));
1.1 kristaps 360:
1.49 schwarze 361: memcpy(&arg->argv[(int)arg->argc - 1], &tmp,
362: sizeof(struct mdoc_argv));
1.1 kristaps 363:
364: return(ARGV_ARG);
365: }
366:
367: void
368: mdoc_argv_free(struct mdoc_arg *p)
369: {
1.23 schwarze 370: int i;
1.1 kristaps 371:
372: if (NULL == p)
373: return;
374:
375: if (p->refcnt) {
376: --(p->refcnt);
377: if (p->refcnt)
378: return;
379: }
380: assert(p->argc);
381:
1.24 schwarze 382: for (i = (int)p->argc - 1; i >= 0; i--)
1.37 schwarze 383: argn_free(p, i);
1.23 schwarze 384:
385: free(p->argv);
386: free(p);
387: }
388:
1.37 schwarze 389: static void
390: argn_free(struct mdoc_arg *p, int iarg)
1.23 schwarze 391: {
1.33 schwarze 392: struct mdoc_argv *arg;
1.23 schwarze 393: int j;
1.1 kristaps 394:
1.33 schwarze 395: arg = &p->argv[iarg];
396:
1.23 schwarze 397: if (arg->sz && arg->value) {
1.49 schwarze 398: for (j = (int)arg->sz - 1; j >= 0; j--)
1.23 schwarze 399: free(arg->value[j]);
400: free(arg->value);
1.1 kristaps 401: }
402:
1.23 schwarze 403: for (--p->argc; iarg < (int)p->argc; iarg++)
404: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 405: }
406:
1.24 schwarze 407: enum margserr
1.42 schwarze 408: mdoc_zargs(struct mdoc *mdoc, int line, int *pos, char *buf, char **v)
1.12 schwarze 409: {
410:
1.42 schwarze 411: return(args(mdoc, line, pos, buf, ARGSFL_NONE, v));
1.12 schwarze 412: }
413:
1.24 schwarze 414: enum margserr
1.49 schwarze 415: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.22 schwarze 416: char *buf, enum mdoct tok, char **v)
1.1 kristaps 417: {
1.38 schwarze 418: enum argsflag fl;
1.1 kristaps 419: struct mdoc_node *n;
420:
1.39 schwarze 421: fl = mdocargs[tok].flags;
1.1 kristaps 422:
1.12 schwarze 423: if (MDOC_It != tok)
1.42 schwarze 424: return(args(mdoc, line, pos, buf, fl, v));
1.12 schwarze 425:
1.30 schwarze 426: /*
427: * We know that we're in an `It', so it's reasonable to expect
428: * us to be sitting in a `Bl'. Someday this may not be the case
429: * (if we allow random `It's sitting out there), so provide a
430: * safe fall-back into the default behaviour.
1.1 kristaps 431: */
432:
1.42 schwarze 433: for (n = mdoc->last; n; n = n->parent)
1.30 schwarze 434: if (MDOC_Bl == n->tok)
1.38 schwarze 435: if (LIST_column == n->norm->Bl.type) {
436: fl = ARGSFL_TABSEP;
437: break;
438: }
1.1 kristaps 439:
1.42 schwarze 440: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 441: }
442:
1.24 schwarze 443: static enum margserr
1.49 schwarze 444: args(struct mdoc *mdoc, int line, int *pos,
1.38 schwarze 445: char *buf, enum argsflag fl, char **v)
1.1 kristaps 446: {
1.30 schwarze 447: char *p, *pp;
1.44 schwarze 448: int pairs;
1.30 schwarze 449: enum margserr rc;
1.1 kristaps 450:
1.28 schwarze 451: if ('\0' == buf[*pos]) {
1.42 schwarze 452: if (MDOC_PPHRASE & mdoc->flags)
1.28 schwarze 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: */
1.42 schwarze 459: if (MDOC_PHRASELIT & mdoc->flags)
460: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
1.28 schwarze 461:
1.42 schwarze 462: mdoc->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 463: return(ARGS_EOLN);
1.28 schwarze 464: }
1.1 kristaps 465:
1.36 schwarze 466: *v = &buf[*pos];
1.1 kristaps 467:
1.38 schwarze 468: if (ARGSFL_DELIM == fl)
469: if (args_checkpunct(buf, *pos))
1.36 schwarze 470: return(ARGS_PUNCT);
1.1 kristaps 471:
1.12 schwarze 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: */
478:
1.38 schwarze 479: if (ARGSFL_TABSEP == fl) {
1.12 schwarze 480: /* Scan ahead to tab (can't be escaped). */
481: p = strchr(*v, '\t');
1.28 schwarze 482: pp = NULL;
1.12 schwarze 483:
484: /* Scan ahead to unescaped `Ta'. */
1.49 schwarze 485: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.28 schwarze 486: for (pp = *v; ; pp++) {
487: if (NULL == (pp = strstr(pp, "Ta")))
488: break;
489: if (pp > *v && ' ' != *(pp - 1))
490: continue;
1.30 schwarze 491: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.28 schwarze 492: break;
493: }
1.1 kristaps 494:
1.24 schwarze 495: /* By default, assume a phrase. */
496: rc = ARGS_PHRASE;
497:
1.49 schwarze 498: /*
1.12 schwarze 499: * Adjust new-buffer position to be beyond delimiter
500: * mark (e.g., Ta -> end + 2).
1.1 kristaps 501: */
1.12 schwarze 502: if (p && pp) {
503: *pos += pp < p ? 2 : 1;
1.24 schwarze 504: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.12 schwarze 505: p = pp < p ? pp : p;
506: } else if (p && ! pp) {
1.24 schwarze 507: rc = ARGS_PPHRASE;
1.12 schwarze 508: *pos += 1;
509: } else if (pp && ! p) {
510: p = pp;
511: *pos += 2;
1.25 schwarze 512: } else {
513: rc = ARGS_PEND;
1.12 schwarze 514: p = strchr(*v, 0);
1.25 schwarze 515: }
1.1 kristaps 516:
1.12 schwarze 517: /* Whitespace check for eoln case... */
1.38 schwarze 518: if ('\0' == *p && ' ' == *(p - 1))
1.42 schwarze 519: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
1.12 schwarze 520:
521: *pos += (int)(p - *v);
522:
523: /* Strip delimiter's preceding whitespace. */
524: pp = p - 1;
525: while (pp > *v && ' ' == *pp) {
526: if (pp > *v && '\\' == *(pp - 1))
527: break;
528: pp--;
529: }
530: *(pp + 1) = 0;
1.1 kristaps 531:
1.12 schwarze 532: /* Strip delimiter's proceeding whitespace. */
533: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
534: /* Skip ahead. */ ;
1.1 kristaps 535:
1.24 schwarze 536: return(rc);
1.46 schwarze 537: }
1.1 kristaps 538:
1.46 schwarze 539: /*
1.12 schwarze 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.
1.46 schwarze 542: * NUL-terminate the literal in place.
1.44 schwarze 543: * Collapse pairs of quotes inside quoted literals.
1.12 schwarze 544: * Whitespace is NOT involved in literal termination.
545: */
1.1 kristaps 546:
1.42 schwarze 547: if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
548: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.28 schwarze 549: *v = &buf[++(*pos)];
550:
1.42 schwarze 551: if (MDOC_PPHRASE & mdoc->flags)
552: mdoc->flags |= MDOC_PHRASELIT;
1.1 kristaps 553:
1.44 schwarze 554: pairs = 0;
1.12 schwarze 555: for ( ; buf[*pos]; (*pos)++) {
1.44 schwarze 556: /* Move following text left after quoted quotes. */
557: if (pairs)
558: buf[*pos - pairs] = buf[*pos];
1.12 schwarze 559: if ('\"' != buf[*pos])
560: continue;
1.44 schwarze 561: /* Unquoted quotes end quoted args. */
1.12 schwarze 562: if ('\"' != buf[*pos + 1])
563: break;
1.44 schwarze 564: /* Quoted quotes collapse. */
565: pairs++;
1.12 schwarze 566: (*pos)++;
567: }
1.44 schwarze 568: if (pairs)
569: buf[*pos - pairs] = '\0';
1.1 kristaps 570:
1.28 schwarze 571: if ('\0' == buf[*pos]) {
1.42 schwarze 572: if (MDOC_PPHRASE & mdoc->flags)
1.13 schwarze 573: return(ARGS_QWORD);
1.42 schwarze 574: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
1.12 schwarze 575: return(ARGS_QWORD);
576: }
1.1 kristaps 577:
1.42 schwarze 578: mdoc->flags &= ~MDOC_PHRASELIT;
1.28 schwarze 579: buf[(*pos)++] = '\0';
1.1 kristaps 580:
1.28 schwarze 581: if ('\0' == buf[*pos])
1.12 schwarze 582: return(ARGS_QWORD);
1.1 kristaps 583:
1.12 schwarze 584: while (' ' == buf[*pos])
585: (*pos)++;
1.1 kristaps 586:
1.38 schwarze 587: if ('\0' == buf[*pos])
1.42 schwarze 588: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
1.1 kristaps 589:
1.12 schwarze 590: return(ARGS_QWORD);
1.1 kristaps 591: }
592:
1.38 schwarze 593: p = &buf[*pos];
1.42 schwarze 594: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 595:
1.12 schwarze 596: return(ARGS_WORD);
1.1 kristaps 597: }
598:
1.49 schwarze 599: /*
1.36 schwarze 600: * Check if the string consists only of space-separated closing
601: * delimiters. This is a bit of a dance: the first must be a close
602: * delimiter, but it may be followed by middle delimiters. Arbitrary
603: * whitespace may separate these tokens.
604: */
605: static int
1.38 schwarze 606: args_checkpunct(const char *buf, int i)
1.36 schwarze 607: {
1.38 schwarze 608: int j;
609: char dbuf[DELIMSZ];
1.36 schwarze 610: enum mdelim d;
611:
612: /* First token must be a close-delimiter. */
613:
1.38 schwarze 614: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
615: dbuf[j] = buf[i];
1.1 kristaps 616:
1.36 schwarze 617: if (DELIMSZ == j)
618: return(0);
619:
1.38 schwarze 620: dbuf[j] = '\0';
621: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.36 schwarze 622: return(0);
623:
1.38 schwarze 624: while (' ' == buf[i])
1.36 schwarze 625: i++;
626:
627: /* Remaining must NOT be open/none. */
1.49 schwarze 628:
1.38 schwarze 629: while (buf[i]) {
1.36 schwarze 630: j = 0;
1.38 schwarze 631: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
632: dbuf[j++] = buf[i++];
1.36 schwarze 633:
634: if (DELIMSZ == j)
635: return(0);
636:
1.38 schwarze 637: dbuf[j] = '\0';
638: d = mdoc_isdelim(dbuf);
1.36 schwarze 639: if (DELIM_NONE == d || DELIM_OPEN == d)
640: return(0);
641:
1.38 schwarze 642: while (' ' == buf[i])
1.36 schwarze 643: i++;
644: }
645:
1.38 schwarze 646: return('\0' == buf[i]);
1.36 schwarze 647: }
648:
1.1 kristaps 649: static int
1.49 schwarze 650: argv_multi(struct mdoc *mdoc, int line,
1.1 kristaps 651: struct mdoc_argv *v, int *pos, char *buf)
652: {
1.24 schwarze 653: enum margserr ac;
1.1 kristaps 654: char *p;
655:
656: for (v->sz = 0; ; v->sz++) {
657: if ('-' == buf[*pos])
658: break;
1.42 schwarze 659: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.24 schwarze 660: if (ARGS_ERROR == ac)
1.1 kristaps 661: return(0);
1.24 schwarze 662: else if (ARGS_EOLN == ac)
1.1 kristaps 663: break;
664:
1.19 schwarze 665: if (0 == v->sz % MULTI_STEP)
1.50 ! schwarze 666: v->value = mandoc_reallocarray(v->value,
! 667: v->sz + MULTI_STEP, sizeof(char *));
1.19 schwarze 668:
669: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 670: }
671:
1.4 schwarze 672: return(1);
1.1 kristaps 673: }
674:
675: static int
1.49 schwarze 676: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 677: struct mdoc_argv *v, int *pos, char *buf)
678: {
1.24 schwarze 679: enum margserr ac;
1.1 kristaps 680: char *p;
681:
1.42 schwarze 682: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.24 schwarze 683: if (ARGS_ERROR == ac)
1.1 kristaps 684: return(0);
1.24 schwarze 685: if (ARGS_EOLN == ac)
1.1 kristaps 686: return(1);
687:
688: v->sz = 1;
1.19 schwarze 689: v->value = mandoc_malloc(sizeof(char *));
690: v->value[0] = mandoc_strdup(p);
1.1 kristaps 691:
692: return(1);
693: }
![[BACK]](/icons/back.gif){kind=icon}
Return to mdoc_argv.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc