Annotation of mandoc/mdoc_argv.c, Revision 1.94
1.94 ! schwarze 1: /* $Id: mdoc_argv.c,v 1.93 2014/04/23 21:06:41 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 */
1.92 schwarze 36: #define DELIMSZ 6 /* max possible size of a delimiter */
1.76 kristaps 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);
1.92 schwarze 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);
1.92 schwarze 59: static int argv_multi(struct mdoc *, int,
1.1 kristaps 60: struct mdoc_argv *, int *, char *);
1.92 schwarze 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 */
1.92 schwarze 153: { ARGSFL_NONE, NULL }, /* Ss */
154: { ARGSFL_NONE, NULL }, /* Pp */
1.78 kristaps 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 */
1.92 schwarze 162: { ARGSFL_DELIM, NULL }, /* Ad */
1.78 kristaps 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 */
1.92 schwarze 167: { ARGSFL_DELIM, NULL }, /* Dv */
168: { ARGSFL_DELIM, NULL }, /* Er */
169: { ARGSFL_DELIM, NULL }, /* Ev */
1.78 kristaps 170: { ARGSFL_NONE, args_Ex }, /* Ex */
1.92 schwarze 171: { ARGSFL_DELIM, NULL }, /* Fa */
172: { ARGSFL_NONE, NULL }, /* Fd */
1.78 kristaps 173: { ARGSFL_DELIM, NULL }, /* Fl */
1.92 schwarze 174: { ARGSFL_DELIM, NULL }, /* Fn */
175: { ARGSFL_DELIM, NULL }, /* Ft */
176: { ARGSFL_DELIM, NULL }, /* Ic */
177: { ARGSFL_DELIM, NULL }, /* In */
1.78 kristaps 178: { ARGSFL_DELIM, NULL }, /* Li */
1.92 schwarze 179: { ARGSFL_NONE, NULL }, /* Nd */
180: { ARGSFL_DELIM, NULL }, /* Nm */
1.78 kristaps 181: { ARGSFL_DELIM, NULL }, /* Op */
1.94 ! schwarze 182: { ARGSFL_DELIM, NULL }, /* Ot */
1.78 kristaps 183: { ARGSFL_DELIM, NULL }, /* Pa */
184: { ARGSFL_NONE, args_Ex }, /* Rv */
1.92 schwarze 185: { ARGSFL_DELIM, NULL }, /* St */
1.78 kristaps 186: { ARGSFL_DELIM, NULL }, /* Va */
1.92 schwarze 187: { ARGSFL_DELIM, NULL }, /* Vt */
1.78 kristaps 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 */
1.92 schwarze 205: { ARGSFL_NONE, args_Bf }, /* Bf */
1.78 kristaps 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 */
1.92 schwarze 216: { ARGSFL_DELIM, NULL }, /* Em */
1.78 kristaps 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 */
1.92 schwarze 244: { ARGSFL_NONE, NULL }, /* Fo */
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 */
1.94 ! schwarze 252: { ARGSFL_DELIM, NULL }, /* Fr */
1.78 kristaps 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 */
1.94 ! schwarze 263: { ARGSFL_DELIM, 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 */
1.91 schwarze 270: { ARGSFL_NONE, NULL }, /* ll */
1.78 kristaps 271: };
272:
273:
1.1 kristaps 274: /*
275: * Parse an argument from line text. This comes in the form of -key
276: * [value0...], which may either have a single mandatory value, at least
277: * one mandatory value, an optional single value, or no value.
278: */
1.40 kristaps 279: enum margverr
1.85 schwarze 280: mdoc_argv(struct mdoc *mdoc, int line, enum mdoct tok,
1.1 kristaps 281: struct mdoc_arg **v, int *pos, char *buf)
282: {
283: char *p, sv;
284: struct mdoc_argv tmp;
285: struct mdoc_arg *arg;
1.78 kristaps 286: const enum mdocargt *ap;
1.1 kristaps 287:
1.52 kristaps 288: if ('\0' == buf[*pos])
1.1 kristaps 289: return(ARGV_EOLN);
1.78 kristaps 290: else if (NULL == (ap = mdocargs[tok].argvs))
291: return(ARGV_WORD);
1.80 kristaps 292: else if ('-' != buf[*pos])
293: return(ARGV_WORD);
1.1 kristaps 294:
1.78 kristaps 295: /* Seek to the first unescaped space. */
1.1 kristaps 296:
297: p = &buf[++(*pos)];
298:
299: assert(*pos > 0);
300:
1.78 kristaps 301: for ( ; buf[*pos] ; (*pos)++)
302: if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
303: break;
1.1 kristaps 304:
1.92 schwarze 305: /*
1.78 kristaps 306: * We want to nil-terminate the word to look it up (it's easier
307: * that way). But we may not have a flag, in which case we need
308: * to restore the line as-is. So keep around the stray byte,
309: * which we'll reset upon exiting (if necessary).
310: */
1.1 kristaps 311:
1.92 schwarze 312: if ('\0' != (sv = buf[*pos]))
1.52 kristaps 313: buf[(*pos)++] = '\0';
1.78 kristaps 314:
315: /*
316: * Now look up the word as a flag. Use temporary storage that
317: * we'll copy into the node's flags, if necessary.
318: */
1.1 kristaps 319:
1.68 kristaps 320: memset(&tmp, 0, sizeof(struct mdoc_argv));
1.78 kristaps 321:
1.1 kristaps 322: tmp.line = line;
323: tmp.pos = *pos;
1.78 kristaps 324: tmp.arg = MDOC_ARG_MAX;
1.1 kristaps 325:
1.78 kristaps 326: while (MDOC_ARG_MAX != (tmp.arg = *ap++))
327: if (0 == strcmp(p, mdoc_argnames[tmp.arg]))
328: break;
1.1 kristaps 329:
1.78 kristaps 330: if (MDOC_ARG_MAX == tmp.arg) {
1.92 schwarze 331: /*
1.78 kristaps 332: * The flag was not found.
333: * Restore saved zeroed byte and return as a word.
334: */
1.1 kristaps 335: if (sv)
336: buf[*pos - 1] = sv;
337: return(ARGV_WORD);
338: }
339:
1.78 kristaps 340: /* Read to the next word (the argument). */
341:
1.1 kristaps 342: while (buf[*pos] && ' ' == buf[*pos])
343: (*pos)++;
344:
1.79 kristaps 345: switch (argvflags[tmp.arg]) {
1.92 schwarze 346: case ARGV_SINGLE:
1.85 schwarze 347: if ( ! argv_single(mdoc, line, &tmp, pos, buf))
1.79 kristaps 348: return(ARGV_ERROR);
349: break;
1.92 schwarze 350: case ARGV_MULTI:
1.85 schwarze 351: if ( ! argv_multi(mdoc, line, &tmp, pos, buf))
1.79 kristaps 352: return(ARGV_ERROR);
353: break;
1.92 schwarze 354: case ARGV_NONE:
1.79 kristaps 355: break;
356: }
1.1 kristaps 357:
1.31 kristaps 358: if (NULL == (arg = *v))
1.32 kristaps 359: arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
1.1 kristaps 360:
361: arg->argc++;
1.93 schwarze 362: arg->argv = mandoc_reallocarray(arg->argv,
363: arg->argc, sizeof(struct mdoc_argv));
1.2 kristaps 364:
1.92 schwarze 365: memcpy(&arg->argv[(int)arg->argc - 1], &tmp,
366: sizeof(struct mdoc_argv));
1.1 kristaps 367:
368: return(ARGV_ARG);
369: }
370:
371: void
372: mdoc_argv_free(struct mdoc_arg *p)
373: {
1.37 kristaps 374: int i;
1.1 kristaps 375:
376: if (NULL == p)
377: return;
378:
379: if (p->refcnt) {
380: --(p->refcnt);
381: if (p->refcnt)
382: return;
383: }
384: assert(p->argc);
385:
1.37 kristaps 386: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 387: argn_free(p, i);
1.37 kristaps 388:
389: free(p->argv);
390: free(p);
391: }
392:
1.71 kristaps 393: static void
394: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 395: {
1.58 kristaps 396: struct mdoc_argv *arg;
1.37 kristaps 397: int j;
1.58 kristaps 398:
399: arg = &p->argv[iarg];
1.1 kristaps 400:
1.37 kristaps 401: if (arg->sz && arg->value) {
1.92 schwarze 402: for (j = (int)arg->sz - 1; j >= 0; j--)
1.37 kristaps 403: free(arg->value[j]);
404: free(arg->value);
1.1 kristaps 405: }
406:
1.37 kristaps 407: for (--p->argc; iarg < (int)p->argc; iarg++)
408: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 409: }
410:
1.39 kristaps 411: enum margserr
1.85 schwarze 412: mdoc_zargs(struct mdoc *mdoc, int line, int *pos, char *buf, char **v)
1.19 kristaps 413: {
414:
1.85 schwarze 415: return(args(mdoc, line, pos, buf, ARGSFL_NONE, v));
1.19 kristaps 416: }
417:
1.39 kristaps 418: enum margserr
1.92 schwarze 419: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.35 kristaps 420: char *buf, enum mdoct tok, char **v)
1.1 kristaps 421: {
1.76 kristaps 422: enum argsflag fl;
1.1 kristaps 423: struct mdoc_node *n;
424:
1.78 kristaps 425: fl = mdocargs[tok].flags;
1.1 kristaps 426:
1.17 kristaps 427: if (MDOC_It != tok)
1.85 schwarze 428: return(args(mdoc, line, pos, buf, fl, v));
1.17 kristaps 429:
1.51 kristaps 430: /*
431: * We know that we're in an `It', so it's reasonable to expect
432: * us to be sitting in a `Bl'. Someday this may not be the case
433: * (if we allow random `It's sitting out there), so provide a
434: * safe fall-back into the default behaviour.
435: */
436:
1.85 schwarze 437: for (n = mdoc->last; n; n = n->parent)
1.50 kristaps 438: if (MDOC_Bl == n->tok)
1.76 kristaps 439: if (LIST_column == n->norm->Bl.type) {
440: fl = ARGSFL_TABSEP;
441: break;
442: }
1.1 kristaps 443:
1.85 schwarze 444: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 445: }
446:
1.39 kristaps 447: static enum margserr
1.92 schwarze 448: args(struct mdoc *mdoc, int line, int *pos,
1.76 kristaps 449: char *buf, enum argsflag fl, char **v)
1.1 kristaps 450: {
1.52 kristaps 451: char *p, *pp;
1.87 schwarze 452: int pairs;
1.52 kristaps 453: enum margserr rc;
1.1 kristaps 454:
1.46 kristaps 455: if ('\0' == buf[*pos]) {
1.85 schwarze 456: if (MDOC_PPHRASE & mdoc->flags)
1.46 kristaps 457: return(ARGS_EOLN);
458: /*
459: * If we're not in a partial phrase and the flag for
460: * being a phrase literal is still set, the punctuation
461: * is unterminated.
462: */
1.85 schwarze 463: if (MDOC_PHRASELIT & mdoc->flags)
464: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
1.46 kristaps 465:
1.85 schwarze 466: mdoc->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 467: return(ARGS_EOLN);
1.46 kristaps 468: }
1.1 kristaps 469:
1.64 kristaps 470: *v = &buf[*pos];
1.17 kristaps 471:
1.76 kristaps 472: if (ARGSFL_DELIM == fl)
1.77 kristaps 473: if (args_checkpunct(buf, *pos))
1.76 kristaps 474: return(ARGS_PUNCT);
1.1 kristaps 475:
1.17 kristaps 476: /*
477: * First handle TABSEP items, restricted to `Bl -column'. This
478: * ignores conventional token parsing and instead uses tabs or
479: * `Ta' macros to separate phrases. Phrases are parsed again
480: * for arguments at a later phase.
481: */
1.1 kristaps 482:
1.76 kristaps 483: if (ARGSFL_TABSEP == fl) {
1.20 kristaps 484: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 485: p = strchr(*v, '\t');
1.46 kristaps 486: pp = NULL;
1.17 kristaps 487:
488: /* Scan ahead to unescaped `Ta'. */
1.92 schwarze 489: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 490: for (pp = *v; ; pp++) {
491: if (NULL == (pp = strstr(pp, "Ta")))
492: break;
493: if (pp > *v && ' ' != *(pp - 1))
494: continue;
1.51 kristaps 495: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 496: break;
497: }
1.17 kristaps 498:
1.42 kristaps 499: /* By default, assume a phrase. */
500: rc = ARGS_PHRASE;
501:
1.92 schwarze 502: /*
1.21 kristaps 503: * Adjust new-buffer position to be beyond delimiter
504: * mark (e.g., Ta -> end + 2).
505: */
1.20 kristaps 506: if (p && pp) {
507: *pos += pp < p ? 2 : 1;
1.42 kristaps 508: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 509: p = pp < p ? pp : p;
510: } else if (p && ! pp) {
1.42 kristaps 511: rc = ARGS_PPHRASE;
1.20 kristaps 512: *pos += 1;
513: } else if (pp && ! p) {
1.17 kristaps 514: p = pp;
1.20 kristaps 515: *pos += 2;
1.44 kristaps 516: } else {
517: rc = ARGS_PEND;
1.20 kristaps 518: p = strchr(*v, 0);
1.44 kristaps 519: }
1.20 kristaps 520:
1.21 kristaps 521: /* Whitespace check for eoln case... */
1.76 kristaps 522: if ('\0' == *p && ' ' == *(p - 1))
1.85 schwarze 523: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
1.20 kristaps 524:
525: *pos += (int)(p - *v);
1.17 kristaps 526:
527: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 528: pp = p - 1;
529: while (pp > *v && ' ' == *pp) {
530: if (pp > *v && '\\' == *(pp - 1))
531: break;
532: pp--;
1.17 kristaps 533: }
1.20 kristaps 534: *(pp + 1) = 0;
1.17 kristaps 535:
1.20 kristaps 536: /* Strip delimiter's proceeding whitespace. */
537: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
538: /* Skip ahead. */ ;
1.1 kristaps 539:
1.41 kristaps 540: return(rc);
1.89 schwarze 541: }
1.1 kristaps 542:
1.89 schwarze 543: /*
1.17 kristaps 544: * Process a quoted literal. A quote begins with a double-quote
545: * and ends with a double-quote NOT preceded by a double-quote.
1.89 schwarze 546: * NUL-terminate the literal in place.
1.87 schwarze 547: * Collapse pairs of quotes inside quoted literals.
1.17 kristaps 548: * Whitespace is NOT involved in literal termination.
549: */
1.1 kristaps 550:
1.85 schwarze 551: if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
552: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 553: *v = &buf[++(*pos)];
554:
1.85 schwarze 555: if (MDOC_PPHRASE & mdoc->flags)
556: mdoc->flags |= MDOC_PHRASELIT;
1.1 kristaps 557:
1.87 schwarze 558: pairs = 0;
1.17 kristaps 559: for ( ; buf[*pos]; (*pos)++) {
1.87 schwarze 560: /* Move following text left after quoted quotes. */
561: if (pairs)
562: buf[*pos - pairs] = buf[*pos];
1.17 kristaps 563: if ('\"' != buf[*pos])
564: continue;
1.87 schwarze 565: /* Unquoted quotes end quoted args. */
1.17 kristaps 566: if ('\"' != buf[*pos + 1])
567: break;
1.87 schwarze 568: /* Quoted quotes collapse. */
569: pairs++;
1.17 kristaps 570: (*pos)++;
571: }
1.87 schwarze 572: if (pairs)
573: buf[*pos - pairs] = '\0';
1.1 kristaps 574:
1.46 kristaps 575: if ('\0' == buf[*pos]) {
1.85 schwarze 576: if (MDOC_PPHRASE & mdoc->flags)
1.23 kristaps 577: return(ARGS_QWORD);
1.85 schwarze 578: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_BADQUOTE);
1.18 kristaps 579: return(ARGS_QWORD);
1.17 kristaps 580: }
1.1 kristaps 581:
1.85 schwarze 582: mdoc->flags &= ~MDOC_PHRASELIT;
1.46 kristaps 583: buf[(*pos)++] = '\0';
1.1 kristaps 584:
1.46 kristaps 585: if ('\0' == buf[*pos])
1.17 kristaps 586: return(ARGS_QWORD);
1.1 kristaps 587:
1.17 kristaps 588: while (' ' == buf[*pos])
589: (*pos)++;
1.1 kristaps 590:
1.76 kristaps 591: if ('\0' == buf[*pos])
1.85 schwarze 592: mdoc_pmsg(mdoc, line, *pos, MANDOCERR_EOLNSPACE);
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:
653: static int
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++) {
661: if ('-' == buf[*pos])
662: break;
1.85 schwarze 663: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 664: if (ARGS_ERROR == ac)
1.1 kristaps 665: return(0);
1.43 kristaps 666: else if (ARGS_EOLN == ac)
1.1 kristaps 667: break;
668:
1.31 kristaps 669: if (0 == v->sz % MULTI_STEP)
1.93 schwarze 670: v->value = mandoc_reallocarray(v->value,
671: v->sz + MULTI_STEP, sizeof(char *));
1.31 kristaps 672:
673: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 674: }
675:
1.7 kristaps 676: return(1);
1.1 kristaps 677: }
678:
679: static int
1.92 schwarze 680: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 681: struct mdoc_argv *v, int *pos, char *buf)
682: {
1.43 kristaps 683: enum margserr ac;
1.1 kristaps 684: char *p;
685:
1.85 schwarze 686: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.43 kristaps 687: if (ARGS_ERROR == ac)
1.1 kristaps 688: return(0);
1.43 kristaps 689: if (ARGS_EOLN == ac)
1.1 kristaps 690: return(1);
691:
692: v->sz = 1;
1.31 kristaps 693: v->value = mandoc_malloc(sizeof(char *));
694: v->value[0] = mandoc_strdup(p);
1.1 kristaps 695:
696: return(1);
697: }
CVSweb