Annotation of mandoc/mdoc_argv.c, Revision 1.102
1.102 ! schwarze 1: /* $Id: mdoc_argv.c,v 1.101 2015/04/02 21:36:50 schwarze Exp $ */
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: *
1.101 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.3 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.101 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.3 kristaps 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:
1.101 schwarze 27: #include "mandoc_aux.h"
28: #include "mandoc.h"
29: #include "roff.h"
1.72 kristaps 30: #include "mdoc.h"
1.101 schwarze 31: #include "libmandoc.h"
1.1 kristaps 32: #include "libmdoc.h"
33:
1.68 kristaps 34: #define MULTI_STEP 5 /* pre-allocate argument values */
1.92 schwarze 35: #define DELIMSZ 6 /* max possible size of a delimiter */
1.76 kristaps 36:
37: enum argsflag {
38: ARGSFL_NONE = 0,
39: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
40: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
41: };
42:
43: enum argvflag {
44: ARGV_NONE, /* no args to flag (e.g., -split) */
45: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
1.83 schwarze 46: ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
1.76 kristaps 47: };
1.1 kristaps 48:
1.78 kristaps 49: struct mdocarg {
50: enum argsflag flags;
51: const enum mdocargt *argvs;
52: };
53:
1.80 kristaps 54: static void argn_free(struct mdoc_arg *, int);
1.92 schwarze 55: static enum margserr args(struct mdoc *, int, int *,
1.76 kristaps 56: char *, enum argsflag, char **);
1.77 kristaps 57: static int args_checkpunct(const char *, int);
1.97 schwarze 58: static void argv_multi(struct mdoc *, int,
1.1 kristaps 59: struct mdoc_argv *, int *, char *);
1.97 schwarze 60: static void argv_single(struct mdoc *, int,
1.1 kristaps 61: struct mdoc_argv *, int *, char *);
1.2 kristaps 62:
1.68 kristaps 63: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 64: ARGV_NONE, /* MDOC_Split */
65: ARGV_NONE, /* MDOC_Nosplit */
66: ARGV_NONE, /* MDOC_Ragged */
67: ARGV_NONE, /* MDOC_Unfilled */
68: ARGV_NONE, /* MDOC_Literal */
1.29 kristaps 69: ARGV_SINGLE, /* MDOC_File */
1.83 schwarze 70: ARGV_SINGLE, /* MDOC_Offset */
1.1 kristaps 71: ARGV_NONE, /* MDOC_Bullet */
72: ARGV_NONE, /* MDOC_Dash */
73: ARGV_NONE, /* MDOC_Hyphen */
74: ARGV_NONE, /* MDOC_Item */
75: ARGV_NONE, /* MDOC_Enum */
76: ARGV_NONE, /* MDOC_Tag */
77: ARGV_NONE, /* MDOC_Diag */
78: ARGV_NONE, /* MDOC_Hang */
79: ARGV_NONE, /* MDOC_Ohang */
80: ARGV_NONE, /* MDOC_Inset */
81: ARGV_MULTI, /* MDOC_Column */
1.83 schwarze 82: ARGV_SINGLE, /* MDOC_Width */
1.1 kristaps 83: ARGV_NONE, /* MDOC_Compact */
1.25 kristaps 84: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 85: ARGV_NONE, /* MDOC_Filled */
86: ARGV_NONE, /* MDOC_Words */
87: ARGV_NONE, /* MDOC_Emphasis */
88: ARGV_NONE, /* MDOC_Symbolic */
89: ARGV_NONE /* MDOC_Symbolic */
90: };
91:
1.73 kristaps 92: static const enum mdocargt args_Ex[] = {
93: MDOC_Std,
94: MDOC_ARG_MAX
95: };
96:
97: static const enum mdocargt args_An[] = {
98: MDOC_Split,
99: MDOC_Nosplit,
100: MDOC_ARG_MAX
101: };
102:
103: static const enum mdocargt args_Bd[] = {
104: MDOC_Ragged,
105: MDOC_Unfilled,
106: MDOC_Filled,
107: MDOC_Literal,
108: MDOC_File,
109: MDOC_Offset,
110: MDOC_Compact,
111: MDOC_Centred,
112: MDOC_ARG_MAX
113: };
114:
115: static const enum mdocargt args_Bf[] = {
116: MDOC_Emphasis,
117: MDOC_Literal,
118: MDOC_Symbolic,
119: MDOC_ARG_MAX
120: };
121:
122: static const enum mdocargt args_Bk[] = {
123: MDOC_Words,
124: MDOC_ARG_MAX
125: };
126:
127: static const enum mdocargt args_Bl[] = {
128: MDOC_Bullet,
129: MDOC_Dash,
130: MDOC_Hyphen,
131: MDOC_Item,
132: MDOC_Enum,
133: MDOC_Tag,
134: MDOC_Diag,
135: MDOC_Hang,
136: MDOC_Ohang,
137: MDOC_Inset,
138: MDOC_Column,
139: MDOC_Width,
140: MDOC_Offset,
141: MDOC_Compact,
142: MDOC_Nested,
143: MDOC_ARG_MAX
144: };
145:
1.78 kristaps 146: static const struct mdocarg mdocargs[MDOC_MAX] = {
1.84 schwarze 147: { ARGSFL_DELIM, NULL }, /* Ap */
1.78 kristaps 148: { ARGSFL_NONE, NULL }, /* Dd */
149: { ARGSFL_NONE, NULL }, /* Dt */
150: { ARGSFL_NONE, NULL }, /* Os */
151: { ARGSFL_NONE, NULL }, /* Sh */
1.92 schwarze 152: { ARGSFL_NONE, NULL }, /* Ss */
153: { ARGSFL_NONE, NULL }, /* Pp */
1.78 kristaps 154: { ARGSFL_DELIM, NULL }, /* D1 */
155: { ARGSFL_DELIM, NULL }, /* Dl */
156: { ARGSFL_NONE, args_Bd }, /* Bd */
157: { ARGSFL_NONE, NULL }, /* Ed */
158: { ARGSFL_NONE, args_Bl }, /* Bl */
159: { ARGSFL_NONE, NULL }, /* El */
160: { ARGSFL_NONE, NULL }, /* It */
1.92 schwarze 161: { ARGSFL_DELIM, NULL }, /* Ad */
1.78 kristaps 162: { ARGSFL_DELIM, args_An }, /* An */
163: { ARGSFL_DELIM, NULL }, /* Ar */
1.86 schwarze 164: { ARGSFL_DELIM, NULL }, /* Cd */
1.78 kristaps 165: { ARGSFL_DELIM, NULL }, /* Cm */
1.92 schwarze 166: { ARGSFL_DELIM, NULL }, /* Dv */
167: { ARGSFL_DELIM, NULL }, /* Er */
168: { ARGSFL_DELIM, NULL }, /* Ev */
1.78 kristaps 169: { ARGSFL_NONE, args_Ex }, /* Ex */
1.92 schwarze 170: { ARGSFL_DELIM, NULL }, /* Fa */
171: { ARGSFL_NONE, NULL }, /* Fd */
1.78 kristaps 172: { ARGSFL_DELIM, NULL }, /* Fl */
1.92 schwarze 173: { ARGSFL_DELIM, NULL }, /* Fn */
174: { ARGSFL_DELIM, NULL }, /* Ft */
175: { ARGSFL_DELIM, NULL }, /* Ic */
176: { ARGSFL_DELIM, NULL }, /* In */
1.78 kristaps 177: { ARGSFL_DELIM, NULL }, /* Li */
1.92 schwarze 178: { ARGSFL_NONE, NULL }, /* Nd */
179: { ARGSFL_DELIM, NULL }, /* Nm */
1.78 kristaps 180: { ARGSFL_DELIM, NULL }, /* Op */
1.94 schwarze 181: { ARGSFL_DELIM, NULL }, /* Ot */
1.78 kristaps 182: { ARGSFL_DELIM, NULL }, /* Pa */
183: { ARGSFL_NONE, args_Ex }, /* Rv */
1.92 schwarze 184: { ARGSFL_DELIM, NULL }, /* St */
1.78 kristaps 185: { ARGSFL_DELIM, NULL }, /* Va */
1.92 schwarze 186: { ARGSFL_DELIM, NULL }, /* Vt */
1.78 kristaps 187: { ARGSFL_DELIM, NULL }, /* Xr */
188: { ARGSFL_NONE, NULL }, /* %A */
189: { ARGSFL_NONE, NULL }, /* %B */
190: { ARGSFL_NONE, NULL }, /* %D */
191: { ARGSFL_NONE, NULL }, /* %I */
192: { ARGSFL_NONE, NULL }, /* %J */
193: { ARGSFL_NONE, NULL }, /* %N */
194: { ARGSFL_NONE, NULL }, /* %O */
195: { ARGSFL_NONE, NULL }, /* %P */
196: { ARGSFL_NONE, NULL }, /* %R */
197: { ARGSFL_NONE, NULL }, /* %T */
198: { ARGSFL_NONE, NULL }, /* %V */
199: { ARGSFL_DELIM, NULL }, /* Ac */
200: { ARGSFL_NONE, NULL }, /* Ao */
201: { ARGSFL_DELIM, NULL }, /* Aq */
202: { ARGSFL_DELIM, NULL }, /* At */
203: { ARGSFL_DELIM, NULL }, /* Bc */
1.92 schwarze 204: { ARGSFL_NONE, args_Bf }, /* Bf */
1.78 kristaps 205: { ARGSFL_NONE, NULL }, /* Bo */
206: { ARGSFL_DELIM, NULL }, /* Bq */
207: { ARGSFL_DELIM, NULL }, /* Bsx */
208: { ARGSFL_DELIM, NULL }, /* Bx */
209: { ARGSFL_NONE, NULL }, /* Db */
210: { ARGSFL_DELIM, NULL }, /* Dc */
211: { ARGSFL_NONE, NULL }, /* Do */
212: { ARGSFL_DELIM, NULL }, /* Dq */
213: { ARGSFL_DELIM, NULL }, /* Ec */
214: { ARGSFL_NONE, NULL }, /* Ef */
1.92 schwarze 215: { ARGSFL_DELIM, NULL }, /* Em */
1.78 kristaps 216: { ARGSFL_NONE, NULL }, /* Eo */
217: { ARGSFL_DELIM, NULL }, /* Fx */
218: { ARGSFL_DELIM, NULL }, /* Ms */
219: { ARGSFL_DELIM, NULL }, /* No */
220: { ARGSFL_DELIM, NULL }, /* Ns */
221: { ARGSFL_DELIM, NULL }, /* Nx */
222: { ARGSFL_DELIM, NULL }, /* Ox */
223: { ARGSFL_DELIM, NULL }, /* Pc */
224: { ARGSFL_DELIM, NULL }, /* Pf */
225: { ARGSFL_NONE, NULL }, /* Po */
226: { ARGSFL_DELIM, NULL }, /* Pq */
227: { ARGSFL_DELIM, NULL }, /* Qc */
228: { ARGSFL_DELIM, NULL }, /* Ql */
229: { ARGSFL_NONE, NULL }, /* Qo */
230: { ARGSFL_DELIM, NULL }, /* Qq */
231: { ARGSFL_NONE, NULL }, /* Re */
232: { ARGSFL_NONE, NULL }, /* Rs */
233: { ARGSFL_DELIM, NULL }, /* Sc */
234: { ARGSFL_NONE, NULL }, /* So */
235: { ARGSFL_DELIM, NULL }, /* Sq */
236: { ARGSFL_NONE, NULL }, /* Sm */
237: { ARGSFL_DELIM, NULL }, /* Sx */
238: { ARGSFL_DELIM, NULL }, /* Sy */
239: { ARGSFL_DELIM, NULL }, /* Tn */
240: { ARGSFL_DELIM, NULL }, /* Ux */
241: { ARGSFL_DELIM, NULL }, /* Xc */
242: { ARGSFL_NONE, NULL }, /* Xo */
1.92 schwarze 243: { ARGSFL_NONE, NULL }, /* Fo */
244: { ARGSFL_DELIM, NULL }, /* Fc */
1.78 kristaps 245: { ARGSFL_NONE, NULL }, /* Oo */
246: { ARGSFL_DELIM, NULL }, /* Oc */
247: { ARGSFL_NONE, args_Bk }, /* Bk */
248: { ARGSFL_NONE, NULL }, /* Ek */
249: { ARGSFL_NONE, NULL }, /* Bt */
250: { ARGSFL_NONE, NULL }, /* Hf */
1.94 schwarze 251: { ARGSFL_DELIM, NULL }, /* Fr */
1.78 kristaps 252: { ARGSFL_NONE, NULL }, /* Ud */
1.86 schwarze 253: { ARGSFL_DELIM, NULL }, /* Lb */
1.78 kristaps 254: { ARGSFL_NONE, NULL }, /* Lp */
255: { ARGSFL_DELIM, NULL }, /* Lk */
256: { ARGSFL_DELIM, NULL }, /* Mt */
257: { ARGSFL_DELIM, NULL }, /* Brq */
258: { ARGSFL_NONE, NULL }, /* Bro */
259: { ARGSFL_DELIM, NULL }, /* Brc */
260: { ARGSFL_NONE, NULL }, /* %C */
261: { ARGSFL_NONE, NULL }, /* Es */
1.94 schwarze 262: { ARGSFL_DELIM, NULL }, /* En */
1.88 schwarze 263: { ARGSFL_DELIM, NULL }, /* Dx */
1.78 kristaps 264: { ARGSFL_NONE, NULL }, /* %Q */
265: { ARGSFL_NONE, NULL }, /* br */
266: { ARGSFL_NONE, NULL }, /* sp */
267: { ARGSFL_NONE, NULL }, /* %U */
268: { ARGSFL_NONE, NULL }, /* Ta */
1.91 schwarze 269: { ARGSFL_NONE, NULL }, /* ll */
1.78 kristaps 270: };
271:
272:
1.1 kristaps 273: /*
1.98 schwarze 274: * Parse flags and their arguments from the input line.
275: * These come in the form -flag [argument ...].
276: * Some flags take no argument, some one, some multiple.
1.1 kristaps 277: */
1.98 schwarze 278: void
1.102 ! schwarze 279: mdoc_argv(struct mdoc *mdoc, int line, int tok,
1.98 schwarze 280: struct mdoc_arg **reta, int *pos, char *buf)
1.1 kristaps 281: {
1.98 schwarze 282: struct mdoc_argv tmpv;
283: struct mdoc_argv **retv;
284: const enum mdocargt *argtable;
285: char *argname;
286: int ipos, retc;
287: char savechar;
1.1 kristaps 288:
1.98 schwarze 289: *reta = NULL;
1.1 kristaps 290:
1.98 schwarze 291: /* Which flags does this macro support? */
1.1 kristaps 292:
1.98 schwarze 293: argtable = mdocargs[tok].argvs;
294: if (argtable == NULL)
295: return;
1.1 kristaps 296:
1.98 schwarze 297: /* Loop over the flags on the input line. */
298:
299: ipos = *pos;
300: while (buf[ipos] == '-') {
301:
302: /* Seek to the first unescaped space. */
1.1 kristaps 303:
1.98 schwarze 304: for (argname = buf + ++ipos; buf[ipos] != '\0'; ipos++)
305: if (buf[ipos] == ' ' && buf[ipos - 1] != '\\')
306: break;
1.1 kristaps 307:
1.98 schwarze 308: /*
309: * We want to nil-terminate the word to look it up.
310: * But we may not have a flag, in which case we need
311: * to restore the line as-is. So keep around the
312: * stray byte, which we'll reset upon exiting.
313: */
314:
315: if ((savechar = buf[ipos]) != '\0')
316: buf[ipos++] = '\0';
1.1 kristaps 317:
1.92 schwarze 318: /*
1.98 schwarze 319: * Now look up the word as a flag. Use temporary
320: * storage that we'll copy into the node's flags.
1.78 kristaps 321: */
1.1 kristaps 322:
1.98 schwarze 323: while ((tmpv.arg = *argtable++) != MDOC_ARG_MAX)
324: if ( ! strcmp(argname, mdoc_argnames[tmpv.arg]))
325: break;
326:
327: /* If it isn't a flag, restore the saved byte. */
328:
329: if (tmpv.arg == MDOC_ARG_MAX) {
330: if (savechar != '\0')
331: buf[ipos - 1] = savechar;
332: break;
333: }
334:
335: /* Read to the next word (the first argument). */
336:
337: while (buf[ipos] == ' ')
338: ipos++;
339:
340: /* Parse the arguments of the flag. */
1.78 kristaps 341:
1.98 schwarze 342: tmpv.line = line;
1.100 schwarze 343: tmpv.pos = *pos;
1.98 schwarze 344: tmpv.sz = 0;
345: tmpv.value = NULL;
1.1 kristaps 346:
1.98 schwarze 347: switch (argvflags[tmpv.arg]) {
348: case ARGV_SINGLE:
349: argv_single(mdoc, line, &tmpv, &ipos, buf);
350: break;
351: case ARGV_MULTI:
352: argv_multi(mdoc, line, &tmpv, &ipos, buf);
353: break;
354: case ARGV_NONE:
355: break;
356: }
357:
358: /* Append to the return values. */
1.1 kristaps 359:
1.98 schwarze 360: if (*reta == NULL)
361: *reta = mandoc_calloc(1, sizeof(**reta));
1.1 kristaps 362:
1.98 schwarze 363: retc = ++(*reta)->argc;
364: retv = &(*reta)->argv;
365: *retv = mandoc_reallocarray(*retv, retc, sizeof(**retv));
366: memcpy(*retv + retc - 1, &tmpv, sizeof(**retv));
1.2 kristaps 367:
1.98 schwarze 368: /* Prepare for parsing the next flag. */
1.1 kristaps 369:
1.98 schwarze 370: *pos = ipos;
371: argtable = mdocargs[tok].argvs;
372: }
1.1 kristaps 373: }
374:
375: void
376: mdoc_argv_free(struct mdoc_arg *p)
377: {
1.37 kristaps 378: int i;
1.1 kristaps 379:
380: if (NULL == p)
381: return;
382:
383: if (p->refcnt) {
384: --(p->refcnt);
385: if (p->refcnt)
386: return;
387: }
388: assert(p->argc);
389:
1.37 kristaps 390: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 391: argn_free(p, i);
1.37 kristaps 392:
393: free(p->argv);
394: free(p);
395: }
396:
1.71 kristaps 397: static void
398: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 399: {
1.58 kristaps 400: struct mdoc_argv *arg;
1.37 kristaps 401: int j;
1.58 kristaps 402:
403: arg = &p->argv[iarg];
1.1 kristaps 404:
1.37 kristaps 405: if (arg->sz && arg->value) {
1.92 schwarze 406: for (j = (int)arg->sz - 1; j >= 0; j--)
1.37 kristaps 407: free(arg->value[j]);
408: free(arg->value);
1.1 kristaps 409: }
410:
1.37 kristaps 411: for (--p->argc; iarg < (int)p->argc; iarg++)
412: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 413: }
414:
1.39 kristaps 415: enum margserr
1.92 schwarze 416: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.102 ! schwarze 417: char *buf, int tok, char **v)
1.1 kristaps 418: {
1.102 ! schwarze 419: struct roff_node *n;
1.99 schwarze 420: char *v_local;
1.76 kristaps 421: enum argsflag fl;
1.1 kristaps 422:
1.99 schwarze 423: if (v == NULL)
424: v = &v_local;
425: fl = tok == MDOC_MAX ? ARGSFL_NONE : mdocargs[tok].flags;
426: if (tok != MDOC_It)
1.85 schwarze 427: return(args(mdoc, line, pos, buf, fl, v));
1.17 kristaps 428:
1.51 kristaps 429: /*
430: * We know that we're in an `It', so it's reasonable to expect
431: * us to be sitting in a `Bl'. Someday this may not be the case
432: * (if we allow random `It's sitting out there), so provide a
433: * safe fall-back into the default behaviour.
434: */
435:
1.85 schwarze 436: for (n = mdoc->last; n; n = n->parent)
1.50 kristaps 437: if (MDOC_Bl == n->tok)
1.76 kristaps 438: if (LIST_column == n->norm->Bl.type) {
439: fl = ARGSFL_TABSEP;
440: break;
441: }
1.1 kristaps 442:
1.85 schwarze 443: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 444: }
445:
1.39 kristaps 446: static enum margserr
1.92 schwarze 447: args(struct mdoc *mdoc, int line, int *pos,
1.76 kristaps 448: char *buf, enum argsflag fl, char **v)
1.1 kristaps 449: {
1.52 kristaps 450: char *p, *pp;
1.87 schwarze 451: int pairs;
1.52 kristaps 452: enum margserr rc;
1.1 kristaps 453:
1.46 kristaps 454: if ('\0' == buf[*pos]) {
1.85 schwarze 455: if (MDOC_PPHRASE & mdoc->flags)
1.46 kristaps 456: return(ARGS_EOLN);
457: /*
458: * If we're not in a partial phrase and the flag for
459: * being a phrase literal is still set, the punctuation
460: * is unterminated.
461: */
1.85 schwarze 462: if (MDOC_PHRASELIT & mdoc->flags)
1.95 schwarze 463: mandoc_msg(MANDOCERR_ARG_QUOTE,
464: mdoc->parse, line, *pos, NULL);
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))
1.95 schwarze 523: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
524: line, *pos, NULL);
1.20 kristaps 525:
526: *pos += (int)(p - *v);
1.17 kristaps 527:
528: /* Strip delimiter's preceding whitespace. */
1.20 kristaps 529: pp = p - 1;
530: while (pp > *v && ' ' == *pp) {
531: if (pp > *v && '\\' == *(pp - 1))
532: break;
533: pp--;
1.17 kristaps 534: }
1.20 kristaps 535: *(pp + 1) = 0;
1.17 kristaps 536:
1.20 kristaps 537: /* Strip delimiter's proceeding whitespace. */
538: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
539: /* Skip ahead. */ ;
1.1 kristaps 540:
1.41 kristaps 541: return(rc);
1.89 schwarze 542: }
1.1 kristaps 543:
1.89 schwarze 544: /*
1.17 kristaps 545: * Process a quoted literal. A quote begins with a double-quote
546: * and ends with a double-quote NOT preceded by a double-quote.
1.89 schwarze 547: * NUL-terminate the literal in place.
1.87 schwarze 548: * Collapse pairs of quotes inside quoted literals.
1.17 kristaps 549: * Whitespace is NOT involved in literal termination.
550: */
1.1 kristaps 551:
1.85 schwarze 552: if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
553: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 554: *v = &buf[++(*pos)];
555:
1.85 schwarze 556: if (MDOC_PPHRASE & mdoc->flags)
557: mdoc->flags |= MDOC_PHRASELIT;
1.1 kristaps 558:
1.87 schwarze 559: pairs = 0;
1.17 kristaps 560: for ( ; buf[*pos]; (*pos)++) {
1.87 schwarze 561: /* Move following text left after quoted quotes. */
562: if (pairs)
563: buf[*pos - pairs] = buf[*pos];
1.17 kristaps 564: if ('\"' != buf[*pos])
565: continue;
1.87 schwarze 566: /* Unquoted quotes end quoted args. */
1.17 kristaps 567: if ('\"' != buf[*pos + 1])
568: break;
1.87 schwarze 569: /* Quoted quotes collapse. */
570: pairs++;
1.17 kristaps 571: (*pos)++;
572: }
1.87 schwarze 573: if (pairs)
574: buf[*pos - pairs] = '\0';
1.1 kristaps 575:
1.46 kristaps 576: if ('\0' == buf[*pos]) {
1.85 schwarze 577: if (MDOC_PPHRASE & mdoc->flags)
1.23 kristaps 578: return(ARGS_QWORD);
1.95 schwarze 579: mandoc_msg(MANDOCERR_ARG_QUOTE,
580: mdoc->parse, line, *pos, NULL);
1.18 kristaps 581: return(ARGS_QWORD);
1.17 kristaps 582: }
1.1 kristaps 583:
1.85 schwarze 584: mdoc->flags &= ~MDOC_PHRASELIT;
1.46 kristaps 585: buf[(*pos)++] = '\0';
1.1 kristaps 586:
1.46 kristaps 587: if ('\0' == buf[*pos])
1.17 kristaps 588: return(ARGS_QWORD);
1.1 kristaps 589:
1.17 kristaps 590: while (' ' == buf[*pos])
591: (*pos)++;
1.1 kristaps 592:
1.76 kristaps 593: if ('\0' == buf[*pos])
1.95 schwarze 594: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
595: line, *pos, NULL);
1.1 kristaps 596:
1.17 kristaps 597: return(ARGS_QWORD);
1.1 kristaps 598: }
599:
1.75 kristaps 600: p = &buf[*pos];
1.85 schwarze 601: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 602:
1.17 kristaps 603: return(ARGS_WORD);
1.64 kristaps 604: }
605:
1.92 schwarze 606: /*
1.64 kristaps 607: * Check if the string consists only of space-separated closing
1.65 kristaps 608: * delimiters. This is a bit of a dance: the first must be a close
609: * delimiter, but it may be followed by middle delimiters. Arbitrary
610: * whitespace may separate these tokens.
1.64 kristaps 611: */
612: static int
1.77 kristaps 613: args_checkpunct(const char *buf, int i)
1.64 kristaps 614: {
1.69 kristaps 615: int j;
616: char dbuf[DELIMSZ];
1.64 kristaps 617: enum mdelim d;
618:
1.65 kristaps 619: /* First token must be a close-delimiter. */
620:
1.69 kristaps 621: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
622: dbuf[j] = buf[i];
1.65 kristaps 623:
624: if (DELIMSZ == j)
625: return(0);
626:
1.69 kristaps 627: dbuf[j] = '\0';
1.70 kristaps 628: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.64 kristaps 629: return(0);
630:
1.69 kristaps 631: while (' ' == buf[i])
1.65 kristaps 632: i++;
633:
634: /* Remaining must NOT be open/none. */
1.92 schwarze 635:
1.69 kristaps 636: while (buf[i]) {
1.65 kristaps 637: j = 0;
1.69 kristaps 638: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
639: dbuf[j++] = buf[i++];
1.65 kristaps 640:
641: if (DELIMSZ == j)
642: return(0);
643:
1.69 kristaps 644: dbuf[j] = '\0';
1.70 kristaps 645: d = mdoc_isdelim(dbuf);
1.64 kristaps 646: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 647: return(0);
648:
1.69 kristaps 649: while (' ' == buf[i])
1.64 kristaps 650: i++;
651: }
1.69 kristaps 652:
653: return('\0' == buf[i]);
1.1 kristaps 654: }
655:
1.97 schwarze 656: static void
1.92 schwarze 657: argv_multi(struct mdoc *mdoc, int line,
1.1 kristaps 658: struct mdoc_argv *v, int *pos, char *buf)
659: {
1.43 kristaps 660: enum margserr ac;
1.1 kristaps 661: char *p;
662:
663: for (v->sz = 0; ; v->sz++) {
1.97 schwarze 664: if (buf[*pos] == '-')
1.1 kristaps 665: break;
1.85 schwarze 666: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 schwarze 667: if (ac == ARGS_EOLN)
1.1 kristaps 668: break;
669:
1.97 schwarze 670: if (v->sz % MULTI_STEP == 0)
1.93 schwarze 671: v->value = mandoc_reallocarray(v->value,
672: v->sz + MULTI_STEP, sizeof(char *));
1.31 kristaps 673:
674: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 675: }
676: }
677:
1.97 schwarze 678: static void
1.92 schwarze 679: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 680: struct mdoc_argv *v, int *pos, char *buf)
681: {
1.43 kristaps 682: enum margserr ac;
1.1 kristaps 683: char *p;
684:
1.85 schwarze 685: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 schwarze 686: if (ac == ARGS_EOLN)
687: return;
1.1 kristaps 688:
689: v->sz = 1;
1.31 kristaps 690: v->value = mandoc_malloc(sizeof(char *));
691: v->value[0] = mandoc_strdup(p);
1.1 kristaps 692: }
CVSweb