Annotation of mandoc/mdoc_argv.c, Revision 1.97
1.97 ! schwarze 1: /* $OpenBSD$ */
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: #include "config.h"
19:
1.1 kristaps 20: #include <sys/types.h>
21:
22: #include <assert.h>
23: #include <stdlib.h>
24: #include <stdio.h>
25: #include <string.h>
26:
1.72 kristaps 27: #include "mdoc.h"
1.49 kristaps 28: #include "mandoc.h"
1.90 schwarze 29: #include "mandoc_aux.h"
1.1 kristaps 30: #include "libmdoc.h"
1.31 kristaps 31: #include "libmandoc.h"
1.1 kristaps 32:
1.68 kristaps 33: #define MULTI_STEP 5 /* pre-allocate argument values */
1.92 schwarze 34: #define DELIMSZ 6 /* max possible size of a delimiter */
1.76 kristaps 35:
36: enum argsflag {
37: ARGSFL_NONE = 0,
38: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
39: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
40: };
41:
42: enum argvflag {
43: ARGV_NONE, /* no args to flag (e.g., -split) */
44: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
1.83 schwarze 45: ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
1.76 kristaps 46: };
1.1 kristaps 47:
1.78 kristaps 48: struct mdocarg {
49: enum argsflag flags;
50: const enum mdocargt *argvs;
51: };
52:
1.80 kristaps 53: static void argn_free(struct mdoc_arg *, int);
1.92 schwarze 54: static enum margserr args(struct mdoc *, int, int *,
1.76 kristaps 55: char *, enum argsflag, char **);
1.77 kristaps 56: static int args_checkpunct(const char *, int);
1.97 ! schwarze 57: static void argv_multi(struct mdoc *, int,
1.1 kristaps 58: struct mdoc_argv *, int *, char *);
1.97 ! schwarze 59: static void argv_single(struct mdoc *, int,
1.1 kristaps 60: struct mdoc_argv *, int *, char *);
1.2 kristaps 61:
1.68 kristaps 62: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 63: ARGV_NONE, /* MDOC_Split */
64: ARGV_NONE, /* MDOC_Nosplit */
65: ARGV_NONE, /* MDOC_Ragged */
66: ARGV_NONE, /* MDOC_Unfilled */
67: ARGV_NONE, /* MDOC_Literal */
1.29 kristaps 68: ARGV_SINGLE, /* MDOC_File */
1.83 schwarze 69: ARGV_SINGLE, /* MDOC_Offset */
1.1 kristaps 70: ARGV_NONE, /* MDOC_Bullet */
71: ARGV_NONE, /* MDOC_Dash */
72: ARGV_NONE, /* MDOC_Hyphen */
73: ARGV_NONE, /* MDOC_Item */
74: ARGV_NONE, /* MDOC_Enum */
75: ARGV_NONE, /* MDOC_Tag */
76: ARGV_NONE, /* MDOC_Diag */
77: ARGV_NONE, /* MDOC_Hang */
78: ARGV_NONE, /* MDOC_Ohang */
79: ARGV_NONE, /* MDOC_Inset */
80: ARGV_MULTI, /* MDOC_Column */
1.83 schwarze 81: ARGV_SINGLE, /* MDOC_Width */
1.1 kristaps 82: ARGV_NONE, /* MDOC_Compact */
1.25 kristaps 83: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 84: ARGV_NONE, /* MDOC_Filled */
85: ARGV_NONE, /* MDOC_Words */
86: ARGV_NONE, /* MDOC_Emphasis */
87: ARGV_NONE, /* MDOC_Symbolic */
88: ARGV_NONE /* MDOC_Symbolic */
89: };
90:
1.73 kristaps 91: static const enum mdocargt args_Ex[] = {
92: MDOC_Std,
93: MDOC_ARG_MAX
94: };
95:
96: static const enum mdocargt args_An[] = {
97: MDOC_Split,
98: MDOC_Nosplit,
99: MDOC_ARG_MAX
100: };
101:
102: static const enum mdocargt args_Bd[] = {
103: MDOC_Ragged,
104: MDOC_Unfilled,
105: MDOC_Filled,
106: MDOC_Literal,
107: MDOC_File,
108: MDOC_Offset,
109: MDOC_Compact,
110: MDOC_Centred,
111: MDOC_ARG_MAX
112: };
113:
114: static const enum mdocargt args_Bf[] = {
115: MDOC_Emphasis,
116: MDOC_Literal,
117: MDOC_Symbolic,
118: MDOC_ARG_MAX
119: };
120:
121: static const enum mdocargt args_Bk[] = {
122: MDOC_Words,
123: MDOC_ARG_MAX
124: };
125:
126: static const enum mdocargt args_Bl[] = {
127: MDOC_Bullet,
128: MDOC_Dash,
129: MDOC_Hyphen,
130: MDOC_Item,
131: MDOC_Enum,
132: MDOC_Tag,
133: MDOC_Diag,
134: MDOC_Hang,
135: MDOC_Ohang,
136: MDOC_Inset,
137: MDOC_Column,
138: MDOC_Width,
139: MDOC_Offset,
140: MDOC_Compact,
141: MDOC_Nested,
142: MDOC_ARG_MAX
143: };
144:
1.78 kristaps 145: static const struct mdocarg mdocargs[MDOC_MAX] = {
1.84 schwarze 146: { ARGSFL_DELIM, NULL }, /* Ap */
1.78 kristaps 147: { ARGSFL_NONE, NULL }, /* Dd */
148: { ARGSFL_NONE, NULL }, /* Dt */
149: { ARGSFL_NONE, NULL }, /* Os */
150: { ARGSFL_NONE, NULL }, /* Sh */
1.92 schwarze 151: { ARGSFL_NONE, NULL }, /* Ss */
152: { ARGSFL_NONE, NULL }, /* Pp */
1.78 kristaps 153: { ARGSFL_DELIM, NULL }, /* D1 */
154: { ARGSFL_DELIM, NULL }, /* Dl */
155: { ARGSFL_NONE, args_Bd }, /* Bd */
156: { ARGSFL_NONE, NULL }, /* Ed */
157: { ARGSFL_NONE, args_Bl }, /* Bl */
158: { ARGSFL_NONE, NULL }, /* El */
159: { ARGSFL_NONE, NULL }, /* It */
1.92 schwarze 160: { ARGSFL_DELIM, NULL }, /* Ad */
1.78 kristaps 161: { ARGSFL_DELIM, args_An }, /* An */
162: { ARGSFL_DELIM, NULL }, /* Ar */
1.86 schwarze 163: { ARGSFL_DELIM, NULL }, /* Cd */
1.78 kristaps 164: { ARGSFL_DELIM, NULL }, /* Cm */
1.92 schwarze 165: { ARGSFL_DELIM, NULL }, /* Dv */
166: { ARGSFL_DELIM, NULL }, /* Er */
167: { ARGSFL_DELIM, NULL }, /* Ev */
1.78 kristaps 168: { ARGSFL_NONE, args_Ex }, /* Ex */
1.92 schwarze 169: { ARGSFL_DELIM, NULL }, /* Fa */
170: { ARGSFL_NONE, NULL }, /* Fd */
1.78 kristaps 171: { ARGSFL_DELIM, NULL }, /* Fl */
1.92 schwarze 172: { ARGSFL_DELIM, NULL }, /* Fn */
173: { ARGSFL_DELIM, NULL }, /* Ft */
174: { ARGSFL_DELIM, NULL }, /* Ic */
175: { ARGSFL_DELIM, NULL }, /* In */
1.78 kristaps 176: { ARGSFL_DELIM, NULL }, /* Li */
1.92 schwarze 177: { ARGSFL_NONE, NULL }, /* Nd */
178: { ARGSFL_DELIM, NULL }, /* Nm */
1.78 kristaps 179: { ARGSFL_DELIM, NULL }, /* Op */
1.94 schwarze 180: { ARGSFL_DELIM, NULL }, /* Ot */
1.78 kristaps 181: { ARGSFL_DELIM, NULL }, /* Pa */
182: { ARGSFL_NONE, args_Ex }, /* Rv */
1.92 schwarze 183: { ARGSFL_DELIM, NULL }, /* St */
1.78 kristaps 184: { ARGSFL_DELIM, NULL }, /* Va */
1.92 schwarze 185: { ARGSFL_DELIM, NULL }, /* Vt */
1.78 kristaps 186: { ARGSFL_DELIM, NULL }, /* Xr */
187: { ARGSFL_NONE, NULL }, /* %A */
188: { ARGSFL_NONE, NULL }, /* %B */
189: { ARGSFL_NONE, NULL }, /* %D */
190: { ARGSFL_NONE, NULL }, /* %I */
191: { ARGSFL_NONE, NULL }, /* %J */
192: { ARGSFL_NONE, NULL }, /* %N */
193: { ARGSFL_NONE, NULL }, /* %O */
194: { ARGSFL_NONE, NULL }, /* %P */
195: { ARGSFL_NONE, NULL }, /* %R */
196: { ARGSFL_NONE, NULL }, /* %T */
197: { ARGSFL_NONE, NULL }, /* %V */
198: { ARGSFL_DELIM, NULL }, /* Ac */
199: { ARGSFL_NONE, NULL }, /* Ao */
200: { ARGSFL_DELIM, NULL }, /* Aq */
201: { ARGSFL_DELIM, NULL }, /* At */
202: { ARGSFL_DELIM, NULL }, /* Bc */
1.92 schwarze 203: { ARGSFL_NONE, args_Bf }, /* Bf */
1.78 kristaps 204: { ARGSFL_NONE, NULL }, /* Bo */
205: { ARGSFL_DELIM, NULL }, /* Bq */
206: { ARGSFL_DELIM, NULL }, /* Bsx */
207: { ARGSFL_DELIM, NULL }, /* Bx */
208: { ARGSFL_NONE, NULL }, /* Db */
209: { ARGSFL_DELIM, NULL }, /* Dc */
210: { ARGSFL_NONE, NULL }, /* Do */
211: { ARGSFL_DELIM, NULL }, /* Dq */
212: { ARGSFL_DELIM, NULL }, /* Ec */
213: { ARGSFL_NONE, NULL }, /* Ef */
1.92 schwarze 214: { ARGSFL_DELIM, NULL }, /* Em */
1.78 kristaps 215: { ARGSFL_NONE, NULL }, /* Eo */
216: { ARGSFL_DELIM, NULL }, /* Fx */
217: { ARGSFL_DELIM, NULL }, /* Ms */
218: { ARGSFL_DELIM, NULL }, /* No */
219: { ARGSFL_DELIM, NULL }, /* Ns */
220: { ARGSFL_DELIM, NULL }, /* Nx */
221: { ARGSFL_DELIM, NULL }, /* Ox */
222: { ARGSFL_DELIM, NULL }, /* Pc */
223: { ARGSFL_DELIM, NULL }, /* Pf */
224: { ARGSFL_NONE, NULL }, /* Po */
225: { ARGSFL_DELIM, NULL }, /* Pq */
226: { ARGSFL_DELIM, NULL }, /* Qc */
227: { ARGSFL_DELIM, NULL }, /* Ql */
228: { ARGSFL_NONE, NULL }, /* Qo */
229: { ARGSFL_DELIM, NULL }, /* Qq */
230: { ARGSFL_NONE, NULL }, /* Re */
231: { ARGSFL_NONE, NULL }, /* Rs */
232: { ARGSFL_DELIM, NULL }, /* Sc */
233: { ARGSFL_NONE, NULL }, /* So */
234: { ARGSFL_DELIM, NULL }, /* Sq */
235: { ARGSFL_NONE, NULL }, /* Sm */
236: { ARGSFL_DELIM, NULL }, /* Sx */
237: { ARGSFL_DELIM, NULL }, /* Sy */
238: { ARGSFL_DELIM, NULL }, /* Tn */
239: { ARGSFL_DELIM, NULL }, /* Ux */
240: { ARGSFL_DELIM, NULL }, /* Xc */
241: { ARGSFL_NONE, NULL }, /* Xo */
1.92 schwarze 242: { ARGSFL_NONE, NULL }, /* Fo */
243: { ARGSFL_DELIM, NULL }, /* Fc */
1.78 kristaps 244: { ARGSFL_NONE, NULL }, /* Oo */
245: { ARGSFL_DELIM, NULL }, /* Oc */
246: { ARGSFL_NONE, args_Bk }, /* Bk */
247: { ARGSFL_NONE, NULL }, /* Ek */
248: { ARGSFL_NONE, NULL }, /* Bt */
249: { ARGSFL_NONE, NULL }, /* Hf */
1.94 schwarze 250: { ARGSFL_DELIM, NULL }, /* Fr */
1.78 kristaps 251: { ARGSFL_NONE, NULL }, /* Ud */
1.86 schwarze 252: { ARGSFL_DELIM, NULL }, /* Lb */
1.78 kristaps 253: { ARGSFL_NONE, NULL }, /* Lp */
254: { ARGSFL_DELIM, NULL }, /* Lk */
255: { ARGSFL_DELIM, NULL }, /* Mt */
256: { ARGSFL_DELIM, NULL }, /* Brq */
257: { ARGSFL_NONE, NULL }, /* Bro */
258: { ARGSFL_DELIM, NULL }, /* Brc */
259: { ARGSFL_NONE, NULL }, /* %C */
260: { ARGSFL_NONE, NULL }, /* Es */
1.94 schwarze 261: { ARGSFL_DELIM, NULL }, /* En */
1.88 schwarze 262: { ARGSFL_DELIM, NULL }, /* Dx */
1.78 kristaps 263: { ARGSFL_NONE, NULL }, /* %Q */
264: { ARGSFL_NONE, NULL }, /* br */
265: { ARGSFL_NONE, NULL }, /* sp */
266: { ARGSFL_NONE, NULL }, /* %U */
267: { ARGSFL_NONE, NULL }, /* Ta */
1.91 schwarze 268: { ARGSFL_NONE, NULL }, /* ll */
1.78 kristaps 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:
1.92 schwarze 303: /*
1.78 kristaps 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:
1.92 schwarze 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) {
1.92 schwarze 329: /*
1.78 kristaps 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]) {
1.92 schwarze 344: case ARGV_SINGLE:
1.97 ! schwarze 345: argv_single(mdoc, line, &tmp, pos, buf);
1.79 kristaps 346: break;
1.92 schwarze 347: case ARGV_MULTI:
1.97 ! schwarze 348: argv_multi(mdoc, line, &tmp, pos, buf);
1.79 kristaps 349: break;
1.92 schwarze 350: case ARGV_NONE:
1.79 kristaps 351: break;
352: }
1.1 kristaps 353:
1.31 kristaps 354: if (NULL == (arg = *v))
1.32 kristaps 355: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 356:
357: arg->argc++;
1.93 schwarze 358: arg->argv = mandoc_reallocarray(arg->argv,
359: arg->argc, sizeof(struct mdoc_argv));
1.2 kristaps 360:
1.92 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: {
1.37 kristaps 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:
1.37 kristaps 382: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 383: argn_free(p, i);
1.37 kristaps 384:
385: free(p->argv);
386: free(p);
387: }
388:
1.71 kristaps 389: static void
390: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 391: {
1.58 kristaps 392: struct mdoc_argv *arg;
1.37 kristaps 393: int j;
1.58 kristaps 394:
395: arg = &p->argv[iarg];
1.1 kristaps 396:
1.37 kristaps 397: if (arg->sz && arg->value) {
1.92 schwarze 398: for (j = (int)arg->sz - 1; j >= 0; j--)
1.37 kristaps 399: free(arg->value[j]);
400: free(arg->value);
1.1 kristaps 401: }
402:
1.37 kristaps 403: for (--p->argc; iarg < (int)p->argc; iarg++)
404: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 405: }
406:
1.39 kristaps 407: enum margserr
1.85 schwarze 408: mdoc_zargs(struct mdoc *mdoc, int line, int *pos, char *buf, char **v)
1.19 kristaps 409: {
410:
1.85 schwarze 411: return(args(mdoc, line, pos, buf, ARGSFL_NONE, v));
1.19 kristaps 412: }
413:
1.39 kristaps 414: enum margserr
1.92 schwarze 415: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.35 kristaps 416: char *buf, enum mdoct tok, char **v)
1.1 kristaps 417: {
1.76 kristaps 418: enum argsflag fl;
1.1 kristaps 419: struct mdoc_node *n;
420:
1.78 kristaps 421: fl = mdocargs[tok].flags;
1.1 kristaps 422:
1.17 kristaps 423: if (MDOC_It != tok)
1.85 schwarze 424: return(args(mdoc, line, pos, buf, fl, v));
1.17 kristaps 425:
1.51 kristaps 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.
431: */
432:
1.85 schwarze 433: for (n = mdoc->last; n; n = n->parent)
1.50 kristaps 434: if (MDOC_Bl == n->tok)
1.76 kristaps 435: if (LIST_column == n->norm->Bl.type) {
436: fl = ARGSFL_TABSEP;
437: break;
438: }
1.1 kristaps 439:
1.85 schwarze 440: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 441: }
442:
1.39 kristaps 443: static enum margserr
1.92 schwarze 444: args(struct mdoc *mdoc, int line, int *pos,
1.76 kristaps 445: char *buf, enum argsflag fl, char **v)
1.1 kristaps 446: {
1.52 kristaps 447: char *p, *pp;
1.87 schwarze 448: int pairs;
1.52 kristaps 449: enum margserr rc;
1.1 kristaps 450:
1.46 kristaps 451: if ('\0' == buf[*pos]) {
1.85 schwarze 452: if (MDOC_PPHRASE & mdoc->flags)
1.46 kristaps 453: return(ARGS_EOLN);
454: /*
455: * If we're not in a partial phrase and the flag for
456: * being a phrase literal is still set, the punctuation
457: * is unterminated.
458: */
1.85 schwarze 459: if (MDOC_PHRASELIT & mdoc->flags)
1.95 schwarze 460: mandoc_msg(MANDOCERR_ARG_QUOTE,
461: mdoc->parse, line, *pos, NULL);
1.46 kristaps 462:
1.85 schwarze 463: mdoc->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 464: return(ARGS_EOLN);
1.46 kristaps 465: }
1.1 kristaps 466:
1.64 kristaps 467: *v = &buf[*pos];
1.17 kristaps 468:
1.76 kristaps 469: if (ARGSFL_DELIM == fl)
1.77 kristaps 470: if (args_checkpunct(buf, *pos))
1.76 kristaps 471: return(ARGS_PUNCT);
1.1 kristaps 472:
1.17 kristaps 473: /*
474: * First handle TABSEP items, restricted to `Bl -column'. This
475: * ignores conventional token parsing and instead uses tabs or
476: * `Ta' macros to separate phrases. Phrases are parsed again
477: * for arguments at a later phase.
478: */
1.1 kristaps 479:
1.76 kristaps 480: if (ARGSFL_TABSEP == fl) {
1.20 kristaps 481: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 482: p = strchr(*v, '\t');
1.46 kristaps 483: pp = NULL;
1.17 kristaps 484:
485: /* Scan ahead to unescaped `Ta'. */
1.92 schwarze 486: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 487: for (pp = *v; ; pp++) {
488: if (NULL == (pp = strstr(pp, "Ta")))
489: break;
490: if (pp > *v && ' ' != *(pp - 1))
491: continue;
1.51 kristaps 492: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 493: break;
494: }
1.17 kristaps 495:
1.42 kristaps 496: /* By default, assume a phrase. */
497: rc = ARGS_PHRASE;
498:
1.92 schwarze 499: /*
1.21 kristaps 500: * Adjust new-buffer position to be beyond delimiter
501: * mark (e.g., Ta -> end + 2).
502: */
1.20 kristaps 503: if (p && pp) {
504: *pos += pp < p ? 2 : 1;
1.42 kristaps 505: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 506: p = pp < p ? pp : p;
507: } else if (p && ! pp) {
1.42 kristaps 508: rc = ARGS_PPHRASE;
1.20 kristaps 509: *pos += 1;
510: } else if (pp && ! p) {
1.17 kristaps 511: p = pp;
1.20 kristaps 512: *pos += 2;
1.44 kristaps 513: } else {
514: rc = ARGS_PEND;
1.20 kristaps 515: p = strchr(*v, 0);
1.44 kristaps 516: }
1.20 kristaps 517:
1.21 kristaps 518: /* Whitespace check for eoln case... */
1.76 kristaps 519: if ('\0' == *p && ' ' == *(p - 1))
1.95 schwarze 520: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
521: line, *pos, NULL);
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);
1.89 schwarze 539: }
1.1 kristaps 540:
1.89 schwarze 541: /*
1.17 kristaps 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.
1.89 schwarze 544: * NUL-terminate the literal in place.
1.87 schwarze 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);
1.95 schwarze 576: mandoc_msg(MANDOCERR_ARG_QUOTE,
577: mdoc->parse, line, *pos, NULL);
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])
1.95 schwarze 591: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
592: line, *pos, NULL);
1.1 kristaps 593:
1.17 kristaps 594: return(ARGS_QWORD);
1.1 kristaps 595: }
596:
1.75 kristaps 597: p = &buf[*pos];
1.85 schwarze 598: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 599:
1.17 kristaps 600: return(ARGS_WORD);
1.64 kristaps 601: }
602:
1.92 schwarze 603: /*
1.64 kristaps 604: * Check if the string consists only of space-separated closing
1.65 kristaps 605: * delimiters. This is a bit of a dance: the first must be a close
606: * delimiter, but it may be followed by middle delimiters. Arbitrary
607: * whitespace may separate these tokens.
1.64 kristaps 608: */
609: static int
1.77 kristaps 610: args_checkpunct(const char *buf, int i)
1.64 kristaps 611: {
1.69 kristaps 612: int j;
613: char dbuf[DELIMSZ];
1.64 kristaps 614: enum mdelim d;
615:
1.65 kristaps 616: /* First token must be a close-delimiter. */
617:
1.69 kristaps 618: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
619: dbuf[j] = buf[i];
1.65 kristaps 620:
621: if (DELIMSZ == j)
622: return(0);
623:
1.69 kristaps 624: dbuf[j] = '\0';
1.70 kristaps 625: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.64 kristaps 626: return(0);
627:
1.69 kristaps 628: while (' ' == buf[i])
1.65 kristaps 629: i++;
630:
631: /* Remaining must NOT be open/none. */
1.92 schwarze 632:
1.69 kristaps 633: while (buf[i]) {
1.65 kristaps 634: j = 0;
1.69 kristaps 635: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
636: dbuf[j++] = buf[i++];
1.65 kristaps 637:
638: if (DELIMSZ == j)
639: return(0);
640:
1.69 kristaps 641: dbuf[j] = '\0';
1.70 kristaps 642: d = mdoc_isdelim(dbuf);
1.64 kristaps 643: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 644: return(0);
645:
1.69 kristaps 646: while (' ' == buf[i])
1.64 kristaps 647: i++;
648: }
1.69 kristaps 649:
650: return('\0' == buf[i]);
1.1 kristaps 651: }
652:
1.97 ! schwarze 653: static void
1.92 schwarze 654: argv_multi(struct mdoc *mdoc, int line,
1.1 kristaps 655: struct mdoc_argv *v, int *pos, char *buf)
656: {
1.43 kristaps 657: enum margserr ac;
1.1 kristaps 658: char *p;
659:
660: for (v->sz = 0; ; v->sz++) {
1.97 ! schwarze 661: if (buf[*pos] == '-')
1.1 kristaps 662: break;
1.85 schwarze 663: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 ! schwarze 664: if (ac == ARGS_EOLN)
1.1 kristaps 665: break;
666:
1.97 ! schwarze 667: if (v->sz % MULTI_STEP == 0)
1.93 schwarze 668: v->value = mandoc_reallocarray(v->value,
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: }
674:
1.97 ! schwarze 675: static void
1.92 schwarze 676: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 677: struct mdoc_argv *v, int *pos, char *buf)
678: {
1.43 kristaps 679: enum margserr ac;
1.1 kristaps 680: char *p;
681:
1.85 schwarze 682: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 ! schwarze 683: if (ac == ARGS_EOLN)
! 684: return;
1.1 kristaps 685:
686: v->sz = 1;
1.31 kristaps 687: v->value = mandoc_malloc(sizeof(char *));
688: v->value[0] = mandoc_strdup(p);
1.1 kristaps 689: }
CVSweb