Annotation of src/usr.bin/mandoc/mdoc_argv.c, Revision 1.57
1.57 ! schwarze 1: /* $OpenBSD: mdoc_argv.c,v 1.56 2015/02/04 18:58:09 schwarze Exp $ */
1.1 kristaps 2: /*
1.36 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.54 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
1.2 schwarze 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.57 ! schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.2 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.57 ! schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.2 schwarze 13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 17: */
18: #include <sys/types.h>
19:
20: #include <assert.h>
21: #include <stdlib.h>
22: #include <stdio.h>
23: #include <string.h>
24:
1.57 ! schwarze 25: #include "mandoc_aux.h"
! 26: #include "mandoc.h"
! 27: #include "roff.h"
1.37 schwarze 28: #include "mdoc.h"
1.57 ! schwarze 29: #include "libmandoc.h"
1.1 kristaps 30: #include "libmdoc.h"
31:
1.36 schwarze 32: #define MULTI_STEP 5 /* pre-allocate argument values */
1.49 schwarze 33: #define DELIMSZ 6 /* max possible size of a delimiter */
1.38 schwarze 34:
35: enum argsflag {
36: ARGSFL_NONE = 0,
37: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
38: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
39: };
40:
41: enum argvflag {
42: ARGV_NONE, /* no args to flag (e.g., -split) */
43: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
1.40 schwarze 44: ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
1.38 schwarze 45: };
1.1 kristaps 46:
1.39 schwarze 47: struct mdocarg {
48: enum argsflag flags;
49: const enum mdocargt *argvs;
50: };
51:
52: static void argn_free(struct mdoc_arg *, int);
1.49 schwarze 53: static enum margserr args(struct mdoc *, int, int *,
1.38 schwarze 54: char *, enum argsflag, char **);
55: static int args_checkpunct(const char *, int);
1.53 schwarze 56: static void argv_multi(struct mdoc *, int,
1.1 kristaps 57: struct mdoc_argv *, int *, char *);
1.53 schwarze 58: static void argv_single(struct mdoc *, int,
1.1 kristaps 59: struct mdoc_argv *, int *, char *);
60:
1.36 schwarze 61: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 62: ARGV_NONE, /* MDOC_Split */
63: ARGV_NONE, /* MDOC_Nosplit */
64: ARGV_NONE, /* MDOC_Ragged */
65: ARGV_NONE, /* MDOC_Unfilled */
66: ARGV_NONE, /* MDOC_Literal */
1.17 schwarze 67: ARGV_SINGLE, /* MDOC_File */
1.40 schwarze 68: ARGV_SINGLE, /* MDOC_Offset */
1.1 kristaps 69: ARGV_NONE, /* MDOC_Bullet */
70: ARGV_NONE, /* MDOC_Dash */
71: ARGV_NONE, /* MDOC_Hyphen */
72: ARGV_NONE, /* MDOC_Item */
73: ARGV_NONE, /* MDOC_Enum */
74: ARGV_NONE, /* MDOC_Tag */
75: ARGV_NONE, /* MDOC_Diag */
76: ARGV_NONE, /* MDOC_Hang */
77: ARGV_NONE, /* MDOC_Ohang */
78: ARGV_NONE, /* MDOC_Inset */
79: ARGV_MULTI, /* MDOC_Column */
80: ARGV_SINGLE, /* MDOC_Width */
81: ARGV_NONE, /* MDOC_Compact */
1.14 schwarze 82: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 83: ARGV_NONE, /* MDOC_Filled */
84: ARGV_NONE, /* MDOC_Words */
85: ARGV_NONE, /* MDOC_Emphasis */
86: ARGV_NONE, /* MDOC_Symbolic */
87: ARGV_NONE /* MDOC_Symbolic */
88: };
89:
1.37 schwarze 90: static const enum mdocargt args_Ex[] = {
91: MDOC_Std,
92: MDOC_ARG_MAX
93: };
94:
95: static const enum mdocargt args_An[] = {
96: MDOC_Split,
97: MDOC_Nosplit,
98: MDOC_ARG_MAX
99: };
100:
101: static const enum mdocargt args_Bd[] = {
102: MDOC_Ragged,
103: MDOC_Unfilled,
104: MDOC_Filled,
105: MDOC_Literal,
106: MDOC_File,
107: MDOC_Offset,
108: MDOC_Compact,
109: MDOC_Centred,
110: MDOC_ARG_MAX
111: };
112:
113: static const enum mdocargt args_Bf[] = {
114: MDOC_Emphasis,
115: MDOC_Literal,
116: MDOC_Symbolic,
117: MDOC_ARG_MAX
118: };
119:
120: static const enum mdocargt args_Bk[] = {
121: MDOC_Words,
122: MDOC_ARG_MAX
123: };
124:
125: static const enum mdocargt args_Bl[] = {
126: MDOC_Bullet,
127: MDOC_Dash,
128: MDOC_Hyphen,
129: MDOC_Item,
130: MDOC_Enum,
131: MDOC_Tag,
132: MDOC_Diag,
133: MDOC_Hang,
134: MDOC_Ohang,
135: MDOC_Inset,
136: MDOC_Column,
137: MDOC_Width,
138: MDOC_Offset,
139: MDOC_Compact,
140: MDOC_Nested,
141: MDOC_ARG_MAX
142: };
143:
1.39 schwarze 144: static const struct mdocarg mdocargs[MDOC_MAX] = {
1.41 schwarze 145: { ARGSFL_DELIM, NULL }, /* Ap */
1.39 schwarze 146: { ARGSFL_NONE, NULL }, /* Dd */
147: { ARGSFL_NONE, NULL }, /* Dt */
148: { ARGSFL_NONE, NULL }, /* Os */
149: { ARGSFL_NONE, NULL }, /* Sh */
1.49 schwarze 150: { ARGSFL_NONE, NULL }, /* Ss */
151: { ARGSFL_NONE, NULL }, /* Pp */
1.39 schwarze 152: { ARGSFL_DELIM, NULL }, /* D1 */
153: { ARGSFL_DELIM, NULL }, /* Dl */
154: { ARGSFL_NONE, args_Bd }, /* Bd */
155: { ARGSFL_NONE, NULL }, /* Ed */
156: { ARGSFL_NONE, args_Bl }, /* Bl */
157: { ARGSFL_NONE, NULL }, /* El */
158: { ARGSFL_NONE, NULL }, /* It */
1.49 schwarze 159: { ARGSFL_DELIM, NULL }, /* Ad */
1.39 schwarze 160: { ARGSFL_DELIM, args_An }, /* An */
161: { ARGSFL_DELIM, NULL }, /* Ar */
1.43 schwarze 162: { ARGSFL_DELIM, NULL }, /* Cd */
1.39 schwarze 163: { ARGSFL_DELIM, NULL }, /* Cm */
1.49 schwarze 164: { ARGSFL_DELIM, NULL }, /* Dv */
165: { ARGSFL_DELIM, NULL }, /* Er */
166: { ARGSFL_DELIM, NULL }, /* Ev */
1.39 schwarze 167: { ARGSFL_NONE, args_Ex }, /* Ex */
1.49 schwarze 168: { ARGSFL_DELIM, NULL }, /* Fa */
169: { ARGSFL_NONE, NULL }, /* Fd */
1.39 schwarze 170: { ARGSFL_DELIM, NULL }, /* Fl */
1.49 schwarze 171: { ARGSFL_DELIM, NULL }, /* Fn */
172: { ARGSFL_DELIM, NULL }, /* Ft */
173: { ARGSFL_DELIM, NULL }, /* Ic */
174: { ARGSFL_DELIM, NULL }, /* In */
1.39 schwarze 175: { ARGSFL_DELIM, NULL }, /* Li */
1.49 schwarze 176: { ARGSFL_NONE, NULL }, /* Nd */
177: { ARGSFL_DELIM, NULL }, /* Nm */
1.39 schwarze 178: { ARGSFL_DELIM, NULL }, /* Op */
1.51 schwarze 179: { ARGSFL_DELIM, NULL }, /* Ot */
1.39 schwarze 180: { ARGSFL_DELIM, NULL }, /* Pa */
181: { ARGSFL_NONE, args_Ex }, /* Rv */
1.49 schwarze 182: { ARGSFL_DELIM, NULL }, /* St */
1.39 schwarze 183: { ARGSFL_DELIM, NULL }, /* Va */
1.49 schwarze 184: { ARGSFL_DELIM, NULL }, /* Vt */
1.39 schwarze 185: { ARGSFL_DELIM, NULL }, /* Xr */
186: { ARGSFL_NONE, NULL }, /* %A */
187: { ARGSFL_NONE, NULL }, /* %B */
188: { ARGSFL_NONE, NULL }, /* %D */
189: { ARGSFL_NONE, NULL }, /* %I */
190: { ARGSFL_NONE, NULL }, /* %J */
191: { ARGSFL_NONE, NULL }, /* %N */
192: { ARGSFL_NONE, NULL }, /* %O */
193: { ARGSFL_NONE, NULL }, /* %P */
194: { ARGSFL_NONE, NULL }, /* %R */
195: { ARGSFL_NONE, NULL }, /* %T */
196: { ARGSFL_NONE, NULL }, /* %V */
197: { ARGSFL_DELIM, NULL }, /* Ac */
198: { ARGSFL_NONE, NULL }, /* Ao */
199: { ARGSFL_DELIM, NULL }, /* Aq */
200: { ARGSFL_DELIM, NULL }, /* At */
201: { ARGSFL_DELIM, NULL }, /* Bc */
1.49 schwarze 202: { ARGSFL_NONE, args_Bf }, /* Bf */
1.39 schwarze 203: { ARGSFL_NONE, NULL }, /* Bo */
204: { ARGSFL_DELIM, NULL }, /* Bq */
205: { ARGSFL_DELIM, NULL }, /* Bsx */
206: { ARGSFL_DELIM, NULL }, /* Bx */
207: { ARGSFL_NONE, NULL }, /* Db */
208: { ARGSFL_DELIM, NULL }, /* Dc */
209: { ARGSFL_NONE, NULL }, /* Do */
210: { ARGSFL_DELIM, NULL }, /* Dq */
211: { ARGSFL_DELIM, NULL }, /* Ec */
212: { ARGSFL_NONE, NULL }, /* Ef */
1.49 schwarze 213: { ARGSFL_DELIM, NULL }, /* Em */
1.39 schwarze 214: { ARGSFL_NONE, NULL }, /* Eo */
215: { ARGSFL_DELIM, NULL }, /* Fx */
216: { ARGSFL_DELIM, NULL }, /* Ms */
217: { ARGSFL_DELIM, NULL }, /* No */
218: { ARGSFL_DELIM, NULL }, /* Ns */
219: { ARGSFL_DELIM, NULL }, /* Nx */
220: { ARGSFL_DELIM, NULL }, /* Ox */
221: { ARGSFL_DELIM, NULL }, /* Pc */
222: { ARGSFL_DELIM, NULL }, /* Pf */
223: { ARGSFL_NONE, NULL }, /* Po */
224: { ARGSFL_DELIM, NULL }, /* Pq */
225: { ARGSFL_DELIM, NULL }, /* Qc */
226: { ARGSFL_DELIM, NULL }, /* Ql */
227: { ARGSFL_NONE, NULL }, /* Qo */
228: { ARGSFL_DELIM, NULL }, /* Qq */
229: { ARGSFL_NONE, NULL }, /* Re */
230: { ARGSFL_NONE, NULL }, /* Rs */
231: { ARGSFL_DELIM, NULL }, /* Sc */
232: { ARGSFL_NONE, NULL }, /* So */
233: { ARGSFL_DELIM, NULL }, /* Sq */
234: { ARGSFL_NONE, NULL }, /* Sm */
235: { ARGSFL_DELIM, NULL }, /* Sx */
236: { ARGSFL_DELIM, NULL }, /* Sy */
237: { ARGSFL_DELIM, NULL }, /* Tn */
238: { ARGSFL_DELIM, NULL }, /* Ux */
239: { ARGSFL_DELIM, NULL }, /* Xc */
240: { ARGSFL_NONE, NULL }, /* Xo */
1.49 schwarze 241: { ARGSFL_NONE, NULL }, /* Fo */
242: { ARGSFL_DELIM, NULL }, /* Fc */
1.39 schwarze 243: { ARGSFL_NONE, NULL }, /* Oo */
244: { ARGSFL_DELIM, NULL }, /* Oc */
245: { ARGSFL_NONE, args_Bk }, /* Bk */
246: { ARGSFL_NONE, NULL }, /* Ek */
247: { ARGSFL_NONE, NULL }, /* Bt */
248: { ARGSFL_NONE, NULL }, /* Hf */
1.51 schwarze 249: { ARGSFL_DELIM, NULL }, /* Fr */
1.39 schwarze 250: { ARGSFL_NONE, NULL }, /* Ud */
1.43 schwarze 251: { ARGSFL_DELIM, NULL }, /* Lb */
1.39 schwarze 252: { ARGSFL_NONE, NULL }, /* Lp */
253: { ARGSFL_DELIM, NULL }, /* Lk */
254: { ARGSFL_DELIM, NULL }, /* Mt */
255: { ARGSFL_DELIM, NULL }, /* Brq */
256: { ARGSFL_NONE, NULL }, /* Bro */
257: { ARGSFL_DELIM, NULL }, /* Brc */
258: { ARGSFL_NONE, NULL }, /* %C */
259: { ARGSFL_NONE, NULL }, /* Es */
1.51 schwarze 260: { ARGSFL_DELIM, NULL }, /* En */
1.45 schwarze 261: { ARGSFL_DELIM, NULL }, /* Dx */
1.39 schwarze 262: { ARGSFL_NONE, NULL }, /* %Q */
263: { ARGSFL_NONE, NULL }, /* br */
264: { ARGSFL_NONE, NULL }, /* sp */
265: { ARGSFL_NONE, NULL }, /* %U */
266: { ARGSFL_NONE, NULL }, /* Ta */
1.48 schwarze 267: { ARGSFL_NONE, NULL }, /* ll */
1.39 schwarze 268: };
269:
270:
1.1 kristaps 271: /*
1.54 schwarze 272: * Parse flags and their arguments from the input line.
273: * These come in the form -flag [argument ...].
274: * Some flags take no argument, some one, some multiple.
1.1 kristaps 275: */
1.54 schwarze 276: void
1.42 schwarze 277: mdoc_argv(struct mdoc *mdoc, int line, enum mdoct tok,
1.54 schwarze 278: struct mdoc_arg **reta, int *pos, char *buf)
1.1 kristaps 279: {
1.54 schwarze 280: struct mdoc_argv tmpv;
281: struct mdoc_argv **retv;
282: const enum mdocargt *argtable;
283: char *argname;
284: int ipos, retc;
285: char savechar;
1.1 kristaps 286:
1.54 schwarze 287: *reta = NULL;
1.1 kristaps 288:
1.54 schwarze 289: /* Which flags does this macro support? */
1.1 kristaps 290:
1.54 schwarze 291: argtable = mdocargs[tok].argvs;
292: if (argtable == NULL)
293: return;
1.1 kristaps 294:
1.54 schwarze 295: /* Loop over the flags on the input line. */
296:
297: ipos = *pos;
298: while (buf[ipos] == '-') {
299:
300: /* Seek to the first unescaped space. */
1.1 kristaps 301:
1.54 schwarze 302: for (argname = buf + ++ipos; buf[ipos] != '\0'; ipos++)
303: if (buf[ipos] == ' ' && buf[ipos - 1] != '\\')
304: break;
1.1 kristaps 305:
1.54 schwarze 306: /*
307: * We want to nil-terminate the word to look it up.
308: * But we may not have a flag, in which case we need
309: * to restore the line as-is. So keep around the
310: * stray byte, which we'll reset upon exiting.
311: */
312:
313: if ((savechar = buf[ipos]) != '\0')
314: buf[ipos++] = '\0';
1.1 kristaps 315:
1.49 schwarze 316: /*
1.54 schwarze 317: * Now look up the word as a flag. Use temporary
318: * storage that we'll copy into the node's flags.
1.39 schwarze 319: */
1.1 kristaps 320:
1.54 schwarze 321: while ((tmpv.arg = *argtable++) != MDOC_ARG_MAX)
322: if ( ! strcmp(argname, mdoc_argnames[tmpv.arg]))
323: break;
324:
325: /* If it isn't a flag, restore the saved byte. */
326:
327: if (tmpv.arg == MDOC_ARG_MAX) {
328: if (savechar != '\0')
329: buf[ipos - 1] = savechar;
330: break;
331: }
332:
333: /* Read to the next word (the first argument). */
334:
335: while (buf[ipos] == ' ')
336: ipos++;
337:
338: /* Parse the arguments of the flag. */
1.39 schwarze 339:
1.54 schwarze 340: tmpv.line = line;
1.56 schwarze 341: tmpv.pos = *pos;
1.54 schwarze 342: tmpv.sz = 0;
343: tmpv.value = NULL;
1.1 kristaps 344:
1.54 schwarze 345: switch (argvflags[tmpv.arg]) {
346: case ARGV_SINGLE:
347: argv_single(mdoc, line, &tmpv, &ipos, buf);
348: break;
349: case ARGV_MULTI:
350: argv_multi(mdoc, line, &tmpv, &ipos, buf);
351: break;
352: case ARGV_NONE:
353: break;
354: }
355:
356: /* Append to the return values. */
1.1 kristaps 357:
1.54 schwarze 358: if (*reta == NULL)
359: *reta = mandoc_calloc(1, sizeof(**reta));
1.1 kristaps 360:
1.54 schwarze 361: retc = ++(*reta)->argc;
362: retv = &(*reta)->argv;
363: *retv = mandoc_reallocarray(*retv, retc, sizeof(**retv));
364: memcpy(*retv + retc - 1, &tmpv, sizeof(**retv));
1.1 kristaps 365:
1.54 schwarze 366: /* Prepare for parsing the next flag. */
1.1 kristaps 367:
1.54 schwarze 368: *pos = ipos;
369: argtable = mdocargs[tok].argvs;
370: }
1.1 kristaps 371: }
372:
373: void
374: mdoc_argv_free(struct mdoc_arg *p)
375: {
1.23 schwarze 376: int i;
1.1 kristaps 377:
378: if (NULL == p)
379: return;
380:
381: if (p->refcnt) {
382: --(p->refcnt);
383: if (p->refcnt)
384: return;
385: }
386: assert(p->argc);
387:
1.24 schwarze 388: for (i = (int)p->argc - 1; i >= 0; i--)
1.37 schwarze 389: argn_free(p, i);
1.23 schwarze 390:
391: free(p->argv);
392: free(p);
393: }
394:
1.37 schwarze 395: static void
396: argn_free(struct mdoc_arg *p, int iarg)
1.23 schwarze 397: {
1.33 schwarze 398: struct mdoc_argv *arg;
1.23 schwarze 399: int j;
1.1 kristaps 400:
1.33 schwarze 401: arg = &p->argv[iarg];
402:
1.23 schwarze 403: if (arg->sz && arg->value) {
1.49 schwarze 404: for (j = (int)arg->sz - 1; j >= 0; j--)
1.23 schwarze 405: free(arg->value[j]);
406: free(arg->value);
1.1 kristaps 407: }
408:
1.23 schwarze 409: for (--p->argc; iarg < (int)p->argc; iarg++)
410: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 411: }
412:
1.24 schwarze 413: enum margserr
1.49 schwarze 414: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.22 schwarze 415: char *buf, enum mdoct tok, char **v)
1.1 kristaps 416: {
1.55 schwarze 417: struct mdoc_node *n;
418: char *v_local;
1.38 schwarze 419: enum argsflag fl;
1.1 kristaps 420:
1.55 schwarze 421: if (v == NULL)
422: v = &v_local;
423: fl = tok == MDOC_MAX ? ARGSFL_NONE : mdocargs[tok].flags;
424: if (tok != MDOC_It)
1.42 schwarze 425: return(args(mdoc, line, pos, buf, fl, v));
1.12 schwarze 426:
1.30 schwarze 427: /*
428: * We know that we're in an `It', so it's reasonable to expect
429: * us to be sitting in a `Bl'. Someday this may not be the case
430: * (if we allow random `It's sitting out there), so provide a
431: * safe fall-back into the default behaviour.
1.1 kristaps 432: */
433:
1.42 schwarze 434: for (n = mdoc->last; n; n = n->parent)
1.30 schwarze 435: if (MDOC_Bl == n->tok)
1.38 schwarze 436: if (LIST_column == n->norm->Bl.type) {
437: fl = ARGSFL_TABSEP;
438: break;
439: }
1.1 kristaps 440:
1.42 schwarze 441: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 442: }
443:
1.24 schwarze 444: static enum margserr
1.49 schwarze 445: args(struct mdoc *mdoc, int line, int *pos,
1.38 schwarze 446: char *buf, enum argsflag fl, char **v)
1.1 kristaps 447: {
1.30 schwarze 448: char *p, *pp;
1.44 schwarze 449: int pairs;
1.30 schwarze 450: enum margserr rc;
1.1 kristaps 451:
1.28 schwarze 452: if ('\0' == buf[*pos]) {
1.42 schwarze 453: if (MDOC_PPHRASE & mdoc->flags)
1.28 schwarze 454: return(ARGS_EOLN);
455: /*
456: * If we're not in a partial phrase and the flag for
457: * being a phrase literal is still set, the punctuation
458: * is unterminated.
459: */
1.42 schwarze 460: if (MDOC_PHRASELIT & mdoc->flags)
1.52 schwarze 461: mandoc_msg(MANDOCERR_ARG_QUOTE,
462: mdoc->parse, line, *pos, NULL);
1.28 schwarze 463:
1.42 schwarze 464: mdoc->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 465: return(ARGS_EOLN);
1.28 schwarze 466: }
1.1 kristaps 467:
1.36 schwarze 468: *v = &buf[*pos];
1.1 kristaps 469:
1.38 schwarze 470: if (ARGSFL_DELIM == fl)
471: if (args_checkpunct(buf, *pos))
1.36 schwarze 472: return(ARGS_PUNCT);
1.1 kristaps 473:
1.12 schwarze 474: /*
475: * First handle TABSEP items, restricted to `Bl -column'. This
476: * ignores conventional token parsing and instead uses tabs or
477: * `Ta' macros to separate phrases. Phrases are parsed again
478: * for arguments at a later phase.
479: */
480:
1.38 schwarze 481: if (ARGSFL_TABSEP == fl) {
1.12 schwarze 482: /* Scan ahead to tab (can't be escaped). */
483: p = strchr(*v, '\t');
1.28 schwarze 484: pp = NULL;
1.12 schwarze 485:
486: /* Scan ahead to unescaped `Ta'. */
1.49 schwarze 487: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.28 schwarze 488: for (pp = *v; ; pp++) {
489: if (NULL == (pp = strstr(pp, "Ta")))
490: break;
491: if (pp > *v && ' ' != *(pp - 1))
492: continue;
1.30 schwarze 493: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.28 schwarze 494: break;
495: }
1.1 kristaps 496:
1.24 schwarze 497: /* By default, assume a phrase. */
498: rc = ARGS_PHRASE;
499:
1.49 schwarze 500: /*
1.12 schwarze 501: * Adjust new-buffer position to be beyond delimiter
502: * mark (e.g., Ta -> end + 2).
1.1 kristaps 503: */
1.12 schwarze 504: if (p && pp) {
505: *pos += pp < p ? 2 : 1;
1.24 schwarze 506: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.12 schwarze 507: p = pp < p ? pp : p;
508: } else if (p && ! pp) {
1.24 schwarze 509: rc = ARGS_PPHRASE;
1.12 schwarze 510: *pos += 1;
511: } else if (pp && ! p) {
512: p = pp;
513: *pos += 2;
1.25 schwarze 514: } else {
515: rc = ARGS_PEND;
1.12 schwarze 516: p = strchr(*v, 0);
1.25 schwarze 517: }
1.1 kristaps 518:
1.12 schwarze 519: /* Whitespace check for eoln case... */
1.38 schwarze 520: if ('\0' == *p && ' ' == *(p - 1))
1.52 schwarze 521: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
522: line, *pos, NULL);
1.12 schwarze 523:
524: *pos += (int)(p - *v);
525:
526: /* Strip delimiter's preceding whitespace. */
527: pp = p - 1;
528: while (pp > *v && ' ' == *pp) {
529: if (pp > *v && '\\' == *(pp - 1))
530: break;
531: pp--;
532: }
533: *(pp + 1) = 0;
1.1 kristaps 534:
1.12 schwarze 535: /* Strip delimiter's proceeding whitespace. */
536: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
537: /* Skip ahead. */ ;
1.1 kristaps 538:
1.24 schwarze 539: return(rc);
1.46 schwarze 540: }
1.1 kristaps 541:
1.46 schwarze 542: /*
1.12 schwarze 543: * Process a quoted literal. A quote begins with a double-quote
544: * and ends with a double-quote NOT preceded by a double-quote.
1.46 schwarze 545: * NUL-terminate the literal in place.
1.44 schwarze 546: * Collapse pairs of quotes inside quoted literals.
1.12 schwarze 547: * Whitespace is NOT involved in literal termination.
548: */
1.1 kristaps 549:
1.42 schwarze 550: if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
551: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.28 schwarze 552: *v = &buf[++(*pos)];
553:
1.42 schwarze 554: if (MDOC_PPHRASE & mdoc->flags)
555: mdoc->flags |= MDOC_PHRASELIT;
1.1 kristaps 556:
1.44 schwarze 557: pairs = 0;
1.12 schwarze 558: for ( ; buf[*pos]; (*pos)++) {
1.44 schwarze 559: /* Move following text left after quoted quotes. */
560: if (pairs)
561: buf[*pos - pairs] = buf[*pos];
1.12 schwarze 562: if ('\"' != buf[*pos])
563: continue;
1.44 schwarze 564: /* Unquoted quotes end quoted args. */
1.12 schwarze 565: if ('\"' != buf[*pos + 1])
566: break;
1.44 schwarze 567: /* Quoted quotes collapse. */
568: pairs++;
1.12 schwarze 569: (*pos)++;
570: }
1.44 schwarze 571: if (pairs)
572: buf[*pos - pairs] = '\0';
1.1 kristaps 573:
1.28 schwarze 574: if ('\0' == buf[*pos]) {
1.42 schwarze 575: if (MDOC_PPHRASE & mdoc->flags)
1.13 schwarze 576: return(ARGS_QWORD);
1.52 schwarze 577: mandoc_msg(MANDOCERR_ARG_QUOTE,
578: mdoc->parse, line, *pos, NULL);
1.12 schwarze 579: return(ARGS_QWORD);
580: }
1.1 kristaps 581:
1.42 schwarze 582: mdoc->flags &= ~MDOC_PHRASELIT;
1.28 schwarze 583: buf[(*pos)++] = '\0';
1.1 kristaps 584:
1.28 schwarze 585: if ('\0' == buf[*pos])
1.12 schwarze 586: return(ARGS_QWORD);
1.1 kristaps 587:
1.12 schwarze 588: while (' ' == buf[*pos])
589: (*pos)++;
1.1 kristaps 590:
1.38 schwarze 591: if ('\0' == buf[*pos])
1.52 schwarze 592: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
593: line, *pos, NULL);
1.1 kristaps 594:
1.12 schwarze 595: return(ARGS_QWORD);
1.1 kristaps 596: }
597:
1.38 schwarze 598: p = &buf[*pos];
1.42 schwarze 599: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 600:
1.12 schwarze 601: return(ARGS_WORD);
1.1 kristaps 602: }
603:
1.49 schwarze 604: /*
1.36 schwarze 605: * Check if the string consists only of space-separated closing
606: * delimiters. This is a bit of a dance: the first must be a close
607: * delimiter, but it may be followed by middle delimiters. Arbitrary
608: * whitespace may separate these tokens.
609: */
610: static int
1.38 schwarze 611: args_checkpunct(const char *buf, int i)
1.36 schwarze 612: {
1.38 schwarze 613: int j;
614: char dbuf[DELIMSZ];
1.36 schwarze 615: enum mdelim d;
616:
617: /* First token must be a close-delimiter. */
618:
1.38 schwarze 619: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
620: dbuf[j] = buf[i];
1.1 kristaps 621:
1.36 schwarze 622: if (DELIMSZ == j)
623: return(0);
624:
1.38 schwarze 625: dbuf[j] = '\0';
626: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.36 schwarze 627: return(0);
628:
1.38 schwarze 629: while (' ' == buf[i])
1.36 schwarze 630: i++;
631:
632: /* Remaining must NOT be open/none. */
1.49 schwarze 633:
1.38 schwarze 634: while (buf[i]) {
1.36 schwarze 635: j = 0;
1.38 schwarze 636: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
637: dbuf[j++] = buf[i++];
1.36 schwarze 638:
639: if (DELIMSZ == j)
640: return(0);
641:
1.38 schwarze 642: dbuf[j] = '\0';
643: d = mdoc_isdelim(dbuf);
1.36 schwarze 644: if (DELIM_NONE == d || DELIM_OPEN == d)
645: return(0);
646:
1.38 schwarze 647: while (' ' == buf[i])
1.36 schwarze 648: i++;
649: }
650:
1.38 schwarze 651: return('\0' == buf[i]);
1.36 schwarze 652: }
653:
1.53 schwarze 654: static void
1.49 schwarze 655: argv_multi(struct mdoc *mdoc, int line,
1.1 kristaps 656: struct mdoc_argv *v, int *pos, char *buf)
657: {
1.24 schwarze 658: enum margserr ac;
1.1 kristaps 659: char *p;
660:
661: for (v->sz = 0; ; v->sz++) {
1.53 schwarze 662: if (buf[*pos] == '-')
1.1 kristaps 663: break;
1.42 schwarze 664: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.53 schwarze 665: if (ac == ARGS_EOLN)
1.1 kristaps 666: break;
667:
1.53 schwarze 668: if (v->sz % MULTI_STEP == 0)
1.50 schwarze 669: v->value = mandoc_reallocarray(v->value,
670: v->sz + MULTI_STEP, sizeof(char *));
1.19 schwarze 671:
672: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 673: }
674: }
675:
1.53 schwarze 676: static void
1.49 schwarze 677: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 678: struct mdoc_argv *v, int *pos, char *buf)
679: {
1.24 schwarze 680: enum margserr ac;
1.1 kristaps 681: char *p;
682:
1.42 schwarze 683: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.53 schwarze 684: if (ac == ARGS_EOLN)
685: return;
1.1 kristaps 686:
687: v->sz = 1;
1.19 schwarze 688: v->value = mandoc_malloc(sizeof(char *));
689: v->value[0] = mandoc_strdup(p);
1.1 kristaps 690: }