Annotation of mandoc/mdoc_argv.c, Revision 1.99
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
1.92 schwarze 415: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.35 kristaps 416: char *buf, enum mdoct tok, char **v)
1.1 kristaps 417: {
1.99 ! schwarze 418: struct mdoc_node *n;
! 419: char *v_local;
1.76 kristaps 420: enum argsflag fl;
1.1 kristaps 421:
1.99 ! schwarze 422: if (v == NULL)
! 423: v = &v_local;
! 424: fl = tok == MDOC_MAX ? ARGSFL_NONE : mdocargs[tok].flags;
! 425: if (tok != MDOC_It)
1.85 schwarze 426: return(args(mdoc, line, pos, buf, fl, v));
1.17 kristaps 427:
1.51 kristaps 428: /*
429: * We know that we're in an `It', so it's reasonable to expect
430: * us to be sitting in a `Bl'. Someday this may not be the case
431: * (if we allow random `It's sitting out there), so provide a
432: * safe fall-back into the default behaviour.
433: */
434:
1.85 schwarze 435: for (n = mdoc->last; n; n = n->parent)
1.50 kristaps 436: if (MDOC_Bl == n->tok)
1.76 kristaps 437: if (LIST_column == n->norm->Bl.type) {
438: fl = ARGSFL_TABSEP;
439: break;
440: }
1.1 kristaps 441:
1.85 schwarze 442: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 443: }
444:
1.39 kristaps 445: static enum margserr
1.92 schwarze 446: args(struct mdoc *mdoc, int line, int *pos,
1.76 kristaps 447: char *buf, enum argsflag fl, char **v)
1.1 kristaps 448: {
1.52 kristaps 449: char *p, *pp;
1.87 schwarze 450: int pairs;
1.52 kristaps 451: enum margserr rc;
1.1 kristaps 452:
1.46 kristaps 453: if ('\0' == buf[*pos]) {
1.85 schwarze 454: if (MDOC_PPHRASE & mdoc->flags)
1.46 kristaps 455: return(ARGS_EOLN);
456: /*
457: * If we're not in a partial phrase and the flag for
458: * being a phrase literal is still set, the punctuation
459: * is unterminated.
460: */
1.85 schwarze 461: if (MDOC_PHRASELIT & mdoc->flags)
1.95 schwarze 462: mandoc_msg(MANDOCERR_ARG_QUOTE,
463: mdoc->parse, line, *pos, NULL);
1.46 kristaps 464:
1.85 schwarze 465: mdoc->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 466: return(ARGS_EOLN);
1.46 kristaps 467: }
1.1 kristaps 468:
1.64 kristaps 469: *v = &buf[*pos];
1.17 kristaps 470:
1.76 kristaps 471: if (ARGSFL_DELIM == fl)
1.77 kristaps 472: if (args_checkpunct(buf, *pos))
1.76 kristaps 473: return(ARGS_PUNCT);
1.1 kristaps 474:
1.17 kristaps 475: /*
476: * First handle TABSEP items, restricted to `Bl -column'. This
477: * ignores conventional token parsing and instead uses tabs or
478: * `Ta' macros to separate phrases. Phrases are parsed again
479: * for arguments at a later phase.
480: */
1.1 kristaps 481:
1.76 kristaps 482: if (ARGSFL_TABSEP == fl) {
1.20 kristaps 483: /* Scan ahead to tab (can't be escaped). */
1.17 kristaps 484: p = strchr(*v, '\t');
1.46 kristaps 485: pp = NULL;
1.17 kristaps 486:
487: /* Scan ahead to unescaped `Ta'. */
1.92 schwarze 488: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.46 kristaps 489: for (pp = *v; ; pp++) {
490: if (NULL == (pp = strstr(pp, "Ta")))
491: break;
492: if (pp > *v && ' ' != *(pp - 1))
493: continue;
1.51 kristaps 494: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.46 kristaps 495: break;
496: }
1.17 kristaps 497:
1.42 kristaps 498: /* By default, assume a phrase. */
499: rc = ARGS_PHRASE;
500:
1.92 schwarze 501: /*
1.21 kristaps 502: * Adjust new-buffer position to be beyond delimiter
503: * mark (e.g., Ta -> end + 2).
504: */
1.20 kristaps 505: if (p && pp) {
506: *pos += pp < p ? 2 : 1;
1.42 kristaps 507: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.20 kristaps 508: p = pp < p ? pp : p;
509: } else if (p && ! pp) {
1.42 kristaps 510: rc = ARGS_PPHRASE;
1.20 kristaps 511: *pos += 1;
512: } else if (pp && ! p) {
1.17 kristaps 513: p = pp;
1.20 kristaps 514: *pos += 2;
1.44 kristaps 515: } else {
516: rc = ARGS_PEND;
1.20 kristaps 517: p = strchr(*v, 0);
1.44 kristaps 518: }
1.20 kristaps 519:
1.21 kristaps 520: /* Whitespace check for eoln case... */
1.76 kristaps 521: if ('\0' == *p && ' ' == *(p - 1))
1.95 schwarze 522: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
523: line, *pos, NULL);
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);
1.95 schwarze 578: mandoc_msg(MANDOCERR_ARG_QUOTE,
579: mdoc->parse, line, *pos, NULL);
1.18 kristaps 580: return(ARGS_QWORD);
1.17 kristaps 581: }
1.1 kristaps 582:
1.85 schwarze 583: mdoc->flags &= ~MDOC_PHRASELIT;
1.46 kristaps 584: buf[(*pos)++] = '\0';
1.1 kristaps 585:
1.46 kristaps 586: if ('\0' == buf[*pos])
1.17 kristaps 587: return(ARGS_QWORD);
1.1 kristaps 588:
1.17 kristaps 589: while (' ' == buf[*pos])
590: (*pos)++;
1.1 kristaps 591:
1.76 kristaps 592: if ('\0' == buf[*pos])
1.95 schwarze 593: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
594: line, *pos, NULL);
1.1 kristaps 595:
1.17 kristaps 596: return(ARGS_QWORD);
1.1 kristaps 597: }
598:
1.75 kristaps 599: p = &buf[*pos];
1.85 schwarze 600: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 601:
1.17 kristaps 602: return(ARGS_WORD);
1.64 kristaps 603: }
604:
1.92 schwarze 605: /*
1.64 kristaps 606: * Check if the string consists only of space-separated closing
1.65 kristaps 607: * delimiters. This is a bit of a dance: the first must be a close
608: * delimiter, but it may be followed by middle delimiters. Arbitrary
609: * whitespace may separate these tokens.
1.64 kristaps 610: */
611: static int
1.77 kristaps 612: args_checkpunct(const char *buf, int i)
1.64 kristaps 613: {
1.69 kristaps 614: int j;
615: char dbuf[DELIMSZ];
1.64 kristaps 616: enum mdelim d;
617:
1.65 kristaps 618: /* First token must be a close-delimiter. */
619:
1.69 kristaps 620: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
621: dbuf[j] = buf[i];
1.65 kristaps 622:
623: if (DELIMSZ == j)
624: return(0);
625:
1.69 kristaps 626: dbuf[j] = '\0';
1.70 kristaps 627: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.64 kristaps 628: return(0);
629:
1.69 kristaps 630: while (' ' == buf[i])
1.65 kristaps 631: i++;
632:
633: /* Remaining must NOT be open/none. */
1.92 schwarze 634:
1.69 kristaps 635: while (buf[i]) {
1.65 kristaps 636: j = 0;
1.69 kristaps 637: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
638: dbuf[j++] = buf[i++];
1.65 kristaps 639:
640: if (DELIMSZ == j)
641: return(0);
642:
1.69 kristaps 643: dbuf[j] = '\0';
1.70 kristaps 644: d = mdoc_isdelim(dbuf);
1.64 kristaps 645: if (DELIM_NONE == d || DELIM_OPEN == d)
1.65 kristaps 646: return(0);
647:
1.69 kristaps 648: while (' ' == buf[i])
1.64 kristaps 649: i++;
650: }
1.69 kristaps 651:
652: return('\0' == buf[i]);
1.1 kristaps 653: }
654:
1.97 schwarze 655: static void
1.92 schwarze 656: argv_multi(struct mdoc *mdoc, int line,
1.1 kristaps 657: struct mdoc_argv *v, int *pos, char *buf)
658: {
1.43 kristaps 659: enum margserr ac;
1.1 kristaps 660: char *p;
661:
662: for (v->sz = 0; ; v->sz++) {
1.97 schwarze 663: if (buf[*pos] == '-')
1.1 kristaps 664: break;
1.85 schwarze 665: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 schwarze 666: if (ac == ARGS_EOLN)
1.1 kristaps 667: break;
668:
1.97 schwarze 669: if (v->sz % MULTI_STEP == 0)
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: }
676:
1.97 schwarze 677: static void
1.92 schwarze 678: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 679: struct mdoc_argv *v, int *pos, char *buf)
680: {
1.43 kristaps 681: enum margserr ac;
1.1 kristaps 682: char *p;
683:
1.85 schwarze 684: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 schwarze 685: if (ac == ARGS_EOLN)
686: return;
1.1 kristaps 687:
688: v->sz = 1;
1.31 kristaps 689: v->value = mandoc_malloc(sizeof(char *));
690: v->value[0] = mandoc_strdup(p);
1.1 kristaps 691: }
CVSweb