Annotation of mandoc/mdoc_argv.c, Revision 1.98
1.97 schwarze 1: /* $OpenBSD$ */
1.1 kristaps 2: /*
1.81 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.98 ! schwarze 4: * Copyright (c) 2012, 2014 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: /*
1.98 ! schwarze 273: * Parse flags and their arguments from the input line.
! 274: * These come in the form -flag [argument ...].
! 275: * Some flags take no argument, some one, some multiple.
1.1 kristaps 276: */
1.98 ! schwarze 277: void
1.85 schwarze 278: mdoc_argv(struct mdoc *mdoc, int line, enum mdoct tok,
1.98 ! schwarze 279: struct mdoc_arg **reta, int *pos, char *buf)
1.1 kristaps 280: {
1.98 ! schwarze 281: struct mdoc_argv tmpv;
! 282: struct mdoc_argv **retv;
! 283: const enum mdocargt *argtable;
! 284: char *argname;
! 285: int ipos, retc;
! 286: char savechar;
1.1 kristaps 287:
1.98 ! schwarze 288: *reta = NULL;
1.1 kristaps 289:
1.98 ! schwarze 290: /* Which flags does this macro support? */
1.1 kristaps 291:
1.98 ! schwarze 292: argtable = mdocargs[tok].argvs;
! 293: if (argtable == NULL)
! 294: return;
1.1 kristaps 295:
1.98 ! schwarze 296: /* Loop over the flags on the input line. */
! 297:
! 298: ipos = *pos;
! 299: while (buf[ipos] == '-') {
! 300:
! 301: /* Seek to the first unescaped space. */
1.1 kristaps 302:
1.98 ! schwarze 303: for (argname = buf + ++ipos; buf[ipos] != '\0'; ipos++)
! 304: if (buf[ipos] == ' ' && buf[ipos - 1] != '\\')
! 305: break;
1.1 kristaps 306:
1.98 ! schwarze 307: /*
! 308: * We want to nil-terminate the word to look it up.
! 309: * But we may not have a flag, in which case we need
! 310: * to restore the line as-is. So keep around the
! 311: * stray byte, which we'll reset upon exiting.
! 312: */
! 313:
! 314: if ((savechar = buf[ipos]) != '\0')
! 315: buf[ipos++] = '\0';
1.1 kristaps 316:
1.92 schwarze 317: /*
1.98 ! schwarze 318: * Now look up the word as a flag. Use temporary
! 319: * storage that we'll copy into the node's flags.
1.78 kristaps 320: */
1.1 kristaps 321:
1.98 ! schwarze 322: while ((tmpv.arg = *argtable++) != MDOC_ARG_MAX)
! 323: if ( ! strcmp(argname, mdoc_argnames[tmpv.arg]))
! 324: break;
! 325:
! 326: /* If it isn't a flag, restore the saved byte. */
! 327:
! 328: if (tmpv.arg == MDOC_ARG_MAX) {
! 329: if (savechar != '\0')
! 330: buf[ipos - 1] = savechar;
! 331: break;
! 332: }
! 333:
! 334: /* Read to the next word (the first argument). */
! 335:
! 336: while (buf[ipos] == ' ')
! 337: ipos++;
! 338:
! 339: /* Parse the arguments of the flag. */
1.78 kristaps 340:
1.98 ! schwarze 341: tmpv.line = line;
! 342: tmpv.pos = ipos;
! 343: tmpv.sz = 0;
! 344: tmpv.value = NULL;
1.1 kristaps 345:
1.98 ! schwarze 346: switch (argvflags[tmpv.arg]) {
! 347: case ARGV_SINGLE:
! 348: argv_single(mdoc, line, &tmpv, &ipos, buf);
! 349: break;
! 350: case ARGV_MULTI:
! 351: argv_multi(mdoc, line, &tmpv, &ipos, buf);
! 352: break;
! 353: case ARGV_NONE:
! 354: break;
! 355: }
! 356:
! 357: /* Append to the return values. */
1.1 kristaps 358:
1.98 ! schwarze 359: if (*reta == NULL)
! 360: *reta = mandoc_calloc(1, sizeof(**reta));
1.1 kristaps 361:
1.98 ! schwarze 362: retc = ++(*reta)->argc;
! 363: retv = &(*reta)->argv;
! 364: *retv = mandoc_reallocarray(*retv, retc, sizeof(**retv));
! 365: memcpy(*retv + retc - 1, &tmpv, sizeof(**retv));
1.2 kristaps 366:
1.98 ! schwarze 367: /* Prepare for parsing the next flag. */
1.1 kristaps 368:
1.98 ! schwarze 369: *pos = ipos;
! 370: argtable = mdocargs[tok].argvs;
! 371: }
1.1 kristaps 372: }
373:
374: void
375: mdoc_argv_free(struct mdoc_arg *p)
376: {
1.37 kristaps 377: int i;
1.1 kristaps 378:
379: if (NULL == p)
380: return;
381:
382: if (p->refcnt) {
383: --(p->refcnt);
384: if (p->refcnt)
385: return;
386: }
387: assert(p->argc);
388:
1.37 kristaps 389: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 390: argn_free(p, i);
1.37 kristaps 391:
392: free(p->argv);
393: free(p);
394: }
395:
1.71 kristaps 396: static void
397: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 398: {
1.58 kristaps 399: struct mdoc_argv *arg;
1.37 kristaps 400: int j;
1.58 kristaps 401:
402: arg = &p->argv[iarg];
1.1 kristaps 403:
1.37 kristaps 404: if (arg->sz && arg->value) {
1.92 schwarze 405: for (j = (int)arg->sz - 1; j >= 0; j--)
1.37 kristaps 406: free(arg->value[j]);
407: free(arg->value);
1.1 kristaps 408: }
409:
1.37 kristaps 410: for (--p->argc; iarg < (int)p->argc; iarg++)
411: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 412: }
413:
1.39 kristaps 414: enum margserr
1.85 schwarze 415: mdoc_zargs(struct mdoc *mdoc, int line, int *pos, char *buf, char **v)
1.19 kristaps 416: {
417:
1.85 schwarze 418: return(args(mdoc, line, pos, buf, ARGSFL_NONE, v));
1.19 kristaps 419: }
420:
1.39 kristaps 421: enum margserr
1.92 schwarze 422: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.35 kristaps 423: char *buf, enum mdoct tok, char **v)
1.1 kristaps 424: {
1.76 kristaps 425: enum argsflag fl;
1.1 kristaps 426: struct mdoc_node *n;
427:
1.78 kristaps 428: fl = mdocargs[tok].flags;
1.1 kristaps 429:
1.17 kristaps 430: if (MDOC_It != tok)
1.85 schwarze 431: return(args(mdoc, line, pos, buf, fl, v));
1.17 kristaps 432:
1.51 kristaps 433: /*
434: * We know that we're in an `It', so it's reasonable to expect
435: * us to be sitting in a `Bl'. Someday this may not be the case
436: * (if we allow random `It's sitting out there), so provide a
437: * safe fall-back into the default behaviour.
438: */
439:
1.85 schwarze 440: for (n = mdoc->last; n; n = n->parent)
1.50 kristaps 441: if (MDOC_Bl == n->tok)
1.76 kristaps 442: if (LIST_column == n->norm->Bl.type) {
443: fl = ARGSFL_TABSEP;
444: break;
445: }
1.1 kristaps 446:
1.85 schwarze 447: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 448: }
449:
1.39 kristaps 450: static enum margserr
1.92 schwarze 451: args(struct mdoc *mdoc, int line, int *pos,
1.76 kristaps 452: char *buf, enum argsflag fl, char **v)
1.1 kristaps 453: {
1.52 kristaps 454: char *p, *pp;
1.87 schwarze 455: int pairs;
1.52 kristaps 456: enum margserr rc;
1.1 kristaps 457:
1.46 kristaps 458: if ('\0' == buf[*pos]) {
1.85 schwarze 459: if (MDOC_PPHRASE & mdoc->flags)
1.46 kristaps 460: return(ARGS_EOLN);
461: /*
462: * If we're not in a partial phrase and the flag for
463: * being a phrase literal is still set, the punctuation
464: * is unterminated.
465: */
1.85 schwarze 466: if (MDOC_PHRASELIT & mdoc->flags)
1.95 schwarze 467: mandoc_msg(MANDOCERR_ARG_QUOTE,
468: mdoc->parse, line, *pos, NULL);
1.46 kristaps 469:
1.85 schwarze 470: mdoc->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 471: return(ARGS_EOLN);
1.46 kristaps 472: }
1.1 kristaps 473:
1.64 kristaps 474: *v = &buf[*pos];
1.17 kristaps 475:
1.76 kristaps 476: if (ARGSFL_DELIM == fl)
1.77 kristaps 477: if (args_checkpunct(buf, *pos))
1.76 kristaps 478: return(ARGS_PUNCT);
1.1 kristaps 479:
1.17 kristaps 480: /*
481: * First handle TABSEP items, restricted to `Bl -column'. This
482: * ignores conventional token parsing and instead uses tabs or
483: * `Ta' macros to separate phrases. Phrases are parsed again
484: * for arguments at a later phase.
485: */
1.1 kristaps 486:
1.76 kristaps 487: if (ARGSFL_TABSEP == fl) {
1.20 kristaps 488: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 489: p = strchr(*v, '\t');
1.46 kristaps 490: pp = NULL;
1.17 kristaps 491:
492: /* Scan ahead to unescaped `Ta'. */
1.92 schwarze 493: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 494: for (pp = *v; ; pp++) {
495: if (NULL == (pp = strstr(pp, "Ta")))
496: break;
497: if (pp > *v && ' ' != *(pp - 1))
498: continue;
1.51 kristaps 499: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 500: break;
501: }
1.17 kristaps 502:
1.42 kristaps 503: /* By default, assume a phrase. */
504: rc = ARGS_PHRASE;
505:
1.92 schwarze 506: /*
1.21 kristaps 507: * Adjust new-buffer position to be beyond delimiter
508: * mark (e.g., Ta -> end + 2).
509: */
1.20 kristaps 510: if (p && pp) {
511: *pos += pp < p ? 2 : 1;
1.42 kristaps 512: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 513: p = pp < p ? pp : p;
514: } else if (p && ! pp) {
1.42 kristaps 515: rc = ARGS_PPHRASE;
1.20 kristaps 516: *pos += 1;
517: } else if (pp && ! p) {
1.17 kristaps 518: p = pp;
1.20 kristaps 519: *pos += 2;
1.44 kristaps 520: } else {
521: rc = ARGS_PEND;
1.20 kristaps 522: p = strchr(*v, 0);
1.44 kristaps 523: }
1.20 kristaps 524:
1.21 kristaps 525: /* Whitespace check for eoln case... */
1.76 kristaps 526: if ('\0' == *p && ' ' == *(p - 1))
1.95 schwarze 527: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
528: line, *pos, NULL);
1.20 kristaps 529:
530: *pos += (int)(p - *v);
1.17 kristaps 531:
532: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 533: pp = p - 1;
534: while (pp > *v && ' ' == *pp) {
535: if (pp > *v && '\\' == *(pp - 1))
536: break;
537: pp--;
1.17 kristaps 538: }
1.20 kristaps 539: *(pp + 1) = 0;
1.17 kristaps 540:
1.20 kristaps 541: /* Strip delimiter's proceeding whitespace. */
542: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
543: /* Skip ahead. */ ;
1.1 kristaps 544:
1.41 kristaps 545: return(rc);
1.89 schwarze 546: }
1.1 kristaps 547:
1.89 schwarze 548: /*
1.17 kristaps 549: * Process a quoted literal. A quote begins with a double-quote
550: * and ends with a double-quote NOT preceded by a double-quote.
1.89 schwarze 551: * NUL-terminate the literal in place.
1.87 schwarze 552: * Collapse pairs of quotes inside quoted literals.
1.17 kristaps 553: * Whitespace is NOT involved in literal termination.
554: */
1.1 kristaps 555:
1.85 schwarze 556: if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
557: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 558: *v = &buf[++(*pos)];
559:
1.85 schwarze 560: if (MDOC_PPHRASE & mdoc->flags)
561: mdoc->flags |= MDOC_PHRASELIT;
1.1 kristaps 562:
1.87 schwarze 563: pairs = 0;
1.17 kristaps 564: for ( ; buf[*pos]; (*pos)++) {
1.87 schwarze 565: /* Move following text left after quoted quotes. */
566: if (pairs)
567: buf[*pos - pairs] = buf[*pos];
1.17 kristaps 568: if ('\"' != buf[*pos])
569: continue;
1.87 schwarze 570: /* Unquoted quotes end quoted args. */
1.17 kristaps 571: if ('\"' != buf[*pos + 1])
572: break;
1.87 schwarze 573: /* Quoted quotes collapse. */
574: pairs++;
1.17 kristaps 575: (*pos)++;
576: }
1.87 schwarze 577: if (pairs)
578: buf[*pos - pairs] = '\0';
1.1 kristaps 579:
1.46 kristaps 580: if ('\0' == buf[*pos]) {
1.85 schwarze 581: if (MDOC_PPHRASE & mdoc->flags)
1.23 kristaps 582: return(ARGS_QWORD);
1.95 schwarze 583: mandoc_msg(MANDOCERR_ARG_QUOTE,
584: mdoc->parse, line, *pos, NULL);
1.18 kristaps 585: return(ARGS_QWORD);
1.17 kristaps 586: }
1.1 kristaps 587:
1.85 schwarze 588: mdoc->flags &= ~MDOC_PHRASELIT;
1.46 kristaps 589: buf[(*pos)++] = '\0';
1.1 kristaps 590:
1.46 kristaps 591: if ('\0' == buf[*pos])
1.17 kristaps 592: return(ARGS_QWORD);
1.1 kristaps 593:
1.17 kristaps 594: while (' ' == buf[*pos])
595: (*pos)++;
1.1 kristaps 596:
1.76 kristaps 597: if ('\0' == buf[*pos])
1.95 schwarze 598: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
599: line, *pos, NULL);
1.1 kristaps 600:
1.17 kristaps 601: return(ARGS_QWORD);
1.1 kristaps 602: }
603:
1.75 kristaps 604: p = &buf[*pos];
1.85 schwarze 605: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 606:
1.17 kristaps 607: return(ARGS_WORD);
1.64 kristaps 608: }
609:
1.92 schwarze 610: /*
1.64 kristaps 611: * Check if the string consists only of space-separated closing
1.65 kristaps 612: * delimiters. This is a bit of a dance: the first must be a close
613: * delimiter, but it may be followed by middle delimiters. Arbitrary
614: * whitespace may separate these tokens.
1.64 kristaps 615: */
616: static int
1.77 kristaps 617: args_checkpunct(const char *buf, int i)
1.64 kristaps 618: {
1.69 kristaps 619: int j;
620: char dbuf[DELIMSZ];
1.64 kristaps 621: enum mdelim d;
622:
1.65 kristaps 623: /* First token must be a close-delimiter. */
624:
1.69 kristaps 625: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
626: dbuf[j] = buf[i];
1.65 kristaps 627:
628: if (DELIMSZ == j)
629: return(0);
630:
1.69 kristaps 631: dbuf[j] = '\0';
1.70 kristaps 632: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.64 kristaps 633: return(0);
634:
1.69 kristaps 635: while (' ' == buf[i])
1.65 kristaps 636: i++;
637:
638: /* Remaining must NOT be open/none. */
1.92 schwarze 639:
1.69 kristaps 640: while (buf[i]) {
1.65 kristaps 641: j = 0;
1.69 kristaps 642: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
643: dbuf[j++] = buf[i++];
1.65 kristaps 644:
645: if (DELIMSZ == j)
646: return(0);
647:
1.69 kristaps 648: dbuf[j] = '\0';
1.70 kristaps 649: d = mdoc_isdelim(dbuf);
1.64 kristaps 650: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 651: return(0);
652:
1.69 kristaps 653: while (' ' == buf[i])
1.64 kristaps 654: i++;
655: }
1.69 kristaps 656:
657: return('\0' == buf[i]);
1.1 kristaps 658: }
659:
1.97 schwarze 660: static void
1.92 schwarze 661: argv_multi(struct mdoc *mdoc, int line,
1.1 kristaps 662: struct mdoc_argv *v, int *pos, char *buf)
663: {
1.43 kristaps 664: enum margserr ac;
1.1 kristaps 665: char *p;
666:
667: for (v->sz = 0; ; v->sz++) {
1.97 schwarze 668: if (buf[*pos] == '-')
1.1 kristaps 669: break;
1.85 schwarze 670: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 schwarze 671: if (ac == ARGS_EOLN)
1.1 kristaps 672: break;
673:
1.97 schwarze 674: if (v->sz % MULTI_STEP == 0)
1.93 schwarze 675: v->value = mandoc_reallocarray(v->value,
676: v->sz + MULTI_STEP, sizeof(char *));
1.31 kristaps 677:
678: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 679: }
680: }
681:
1.97 schwarze 682: static void
1.92 schwarze 683: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 684: struct mdoc_argv *v, int *pos, char *buf)
685: {
1.43 kristaps 686: enum margserr ac;
1.1 kristaps 687: char *p;
688:
1.85 schwarze 689: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 schwarze 690: if (ac == ARGS_EOLN)
691: return;
1.1 kristaps 692:
693: v->sz = 1;
1.31 kristaps 694: v->value = mandoc_malloc(sizeof(char *));
695: v->value[0] = mandoc_strdup(p);
1.1 kristaps 696: }
CVSweb