mandoc(1): Update to 1.9.15.
[dragonfly.git] / usr.bin / mandoc / mdoc_argv.c
blob846c55e30dab8cc57199713a5aa01c4ea9913bd1
1 /* $Id: mdoc_argv.c,v 1.33 2010/01/01 17:14:29 kristaps Exp $ */
2 /*
3 * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
5 * Permission to use, copy, modify, and distribute this software for any
6 * purpose with or without fee is hereby granted, provided that the above
7 * copyright notice and this permission notice appear in all copies.
9 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 #include <sys/types.h>
19 #include <assert.h>
20 #include <ctype.h>
21 #include <stdlib.h>
22 #include <stdio.h>
23 #include <string.h>
25 #include "libmdoc.h"
26 #include "libmandoc.h"
29 * Routines to parse arguments of macros. Arguments follow the syntax
30 * of `-arg [val [valN...]]'. Arguments come in all types: quoted
31 * arguments, multiple arguments per value, no-value arguments, etc.
33 * There's no limit to the number or arguments that may be allocated.
36 #define ARGV_NONE (1 << 0)
37 #define ARGV_SINGLE (1 << 1)
38 #define ARGV_MULTI (1 << 2)
39 #define ARGV_OPT_SINGLE (1 << 3)
41 #define MULTI_STEP 5
43 static int argv_a2arg(int, const char *);
44 static int args(struct mdoc *, int, int *,
45 char *, int, char **);
46 static int argv(struct mdoc *, int,
47 struct mdoc_argv *, int *, char *);
48 static int argv_single(struct mdoc *, int,
49 struct mdoc_argv *, int *, char *);
50 static int argv_opt_single(struct mdoc *, int,
51 struct mdoc_argv *, int *, char *);
52 static int argv_multi(struct mdoc *, int,
53 struct mdoc_argv *, int *, char *);
55 /* Per-argument flags. */
57 static int mdoc_argvflags[MDOC_ARG_MAX] = {
58 ARGV_NONE, /* MDOC_Split */
59 ARGV_NONE, /* MDOC_Nosplit */
60 ARGV_NONE, /* MDOC_Ragged */
61 ARGV_NONE, /* MDOC_Unfilled */
62 ARGV_NONE, /* MDOC_Literal */
63 ARGV_SINGLE, /* MDOC_File */
64 ARGV_OPT_SINGLE, /* MDOC_Offset */
65 ARGV_NONE, /* MDOC_Bullet */
66 ARGV_NONE, /* MDOC_Dash */
67 ARGV_NONE, /* MDOC_Hyphen */
68 ARGV_NONE, /* MDOC_Item */
69 ARGV_NONE, /* MDOC_Enum */
70 ARGV_NONE, /* MDOC_Tag */
71 ARGV_NONE, /* MDOC_Diag */
72 ARGV_NONE, /* MDOC_Hang */
73 ARGV_NONE, /* MDOC_Ohang */
74 ARGV_NONE, /* MDOC_Inset */
75 ARGV_MULTI, /* MDOC_Column */
76 ARGV_SINGLE, /* MDOC_Width */
77 ARGV_NONE, /* MDOC_Compact */
78 ARGV_NONE, /* MDOC_Std */
79 ARGV_NONE, /* MDOC_Filled */
80 ARGV_NONE, /* MDOC_Words */
81 ARGV_NONE, /* MDOC_Emphasis */
82 ARGV_NONE, /* MDOC_Symbolic */
83 ARGV_NONE /* MDOC_Symbolic */
86 static int mdoc_argflags[MDOC_MAX] = {
87 0, /* Ap */
88 0, /* Dd */
89 0, /* Dt */
90 0, /* Os */
91 0, /* Sh */
92 0, /* Ss */
93 ARGS_DELIM, /* Pp */
94 ARGS_DELIM, /* D1 */
95 ARGS_DELIM, /* Dl */
96 0, /* Bd */
97 0, /* Ed */
98 0, /* Bl */
99 0, /* El */
100 0, /* It */
101 ARGS_DELIM, /* Ad */
102 ARGS_DELIM, /* An */
103 ARGS_DELIM, /* Ar */
104 0, /* Cd */
105 ARGS_DELIM, /* Cm */
106 ARGS_DELIM, /* Dv */
107 ARGS_DELIM, /* Er */
108 ARGS_DELIM, /* Ev */
109 0, /* Ex */
110 ARGS_DELIM, /* Fa */
111 0, /* Fd */
112 ARGS_DELIM, /* Fl */
113 ARGS_DELIM, /* Fn */
114 ARGS_DELIM, /* Ft */
115 ARGS_DELIM, /* Ic */
116 0, /* In */
117 ARGS_DELIM, /* Li */
118 0, /* Nd */
119 ARGS_DELIM, /* Nm */
120 ARGS_DELIM, /* Op */
121 0, /* Ot */
122 ARGS_DELIM, /* Pa */
123 0, /* Rv */
124 ARGS_DELIM, /* St */
125 ARGS_DELIM, /* Va */
126 ARGS_DELIM, /* Vt */
127 ARGS_DELIM, /* Xr */
128 0, /* %A */
129 0, /* %B */
130 0, /* %D */
131 0, /* %I */
132 0, /* %J */
133 0, /* %N */
134 0, /* %O */
135 0, /* %P */
136 0, /* %R */
137 0, /* %T */
138 0, /* %V */
139 ARGS_DELIM, /* Ac */
140 0, /* Ao */
141 ARGS_DELIM, /* Aq */
142 ARGS_DELIM, /* At */
143 ARGS_DELIM, /* Bc */
144 0, /* Bf */
145 0, /* Bo */
146 ARGS_DELIM, /* Bq */
147 ARGS_DELIM, /* Bsx */
148 ARGS_DELIM, /* Bx */
149 0, /* Db */
150 ARGS_DELIM, /* Dc */
151 0, /* Do */
152 ARGS_DELIM, /* Dq */
153 ARGS_DELIM, /* Ec */
154 0, /* Ef */
155 ARGS_DELIM, /* Em */
156 0, /* Eo */
157 ARGS_DELIM, /* Fx */
158 ARGS_DELIM, /* Ms */
159 ARGS_DELIM, /* No */
160 ARGS_DELIM, /* Ns */
161 ARGS_DELIM, /* Nx */
162 ARGS_DELIM, /* Ox */
163 ARGS_DELIM, /* Pc */
164 ARGS_DELIM, /* Pf */
165 0, /* Po */
166 ARGS_DELIM, /* Pq */
167 ARGS_DELIM, /* Qc */
168 ARGS_DELIM, /* Ql */
169 0, /* Qo */
170 ARGS_DELIM, /* Qq */
171 0, /* Re */
172 0, /* Rs */
173 ARGS_DELIM, /* Sc */
174 0, /* So */
175 ARGS_DELIM, /* Sq */
176 0, /* Sm */
177 ARGS_DELIM, /* Sx */
178 ARGS_DELIM, /* Sy */
179 ARGS_DELIM, /* Tn */
180 ARGS_DELIM, /* Ux */
181 ARGS_DELIM, /* Xc */
182 0, /* Xo */
183 0, /* Fo */
184 0, /* Fc */
185 0, /* Oo */
186 ARGS_DELIM, /* Oc */
187 0, /* Bk */
188 0, /* Ek */
189 0, /* Bt */
190 0, /* Hf */
191 0, /* Fr */
192 0, /* Ud */
193 0, /* Lb */
194 ARGS_DELIM, /* Lp */
195 ARGS_DELIM, /* Lk */
196 ARGS_DELIM, /* Mt */
197 ARGS_DELIM, /* Brq */
198 0, /* Bro */
199 ARGS_DELIM, /* Brc */
200 0, /* %C */
201 0, /* Es */
202 0, /* En */
203 0, /* Dx */
204 0, /* %Q */
205 0, /* br */
206 0, /* sp */
207 0, /* %U */
212 * Parse an argument from line text. This comes in the form of -key
213 * [value0...], which may either have a single mandatory value, at least
214 * one mandatory value, an optional single value, or no value.
217 mdoc_argv(struct mdoc *m, int line, int tok,
218 struct mdoc_arg **v, int *pos, char *buf)
220 char *p, sv;
221 struct mdoc_argv tmp;
222 struct mdoc_arg *arg;
224 if (0 == buf[*pos])
225 return(ARGV_EOLN);
227 assert(' ' != buf[*pos]);
229 /* Parse through to the first unescaped space. */
231 p = &buf[++(*pos)];
233 assert(*pos > 0);
235 /* LINTED */
236 while (buf[*pos]) {
237 if (' ' == buf[*pos])
238 if ('\\' != buf[*pos - 1])
239 break;
240 (*pos)++;
243 /* XXX - save zeroed byte, if not an argument. */
245 sv = 0;
246 if (buf[*pos]) {
247 sv = buf[*pos];
248 buf[(*pos)++] = 0;
251 (void)memset(&tmp, 0, sizeof(struct mdoc_argv));
252 tmp.line = line;
253 tmp.pos = *pos;
255 /* See if our token accepts the argument. */
257 if (MDOC_ARG_MAX == (tmp.arg = argv_a2arg(tok, p))) {
258 /* XXX - restore saved zeroed byte. */
259 if (sv)
260 buf[*pos - 1] = sv;
261 return(ARGV_WORD);
264 while (buf[*pos] && ' ' == buf[*pos])
265 (*pos)++;
267 if ( ! argv(m, line, &tmp, pos, buf))
268 return(ARGV_ERROR);
270 if (NULL == (arg = *v))
271 arg = *v = mandoc_calloc(1, sizeof(struct mdoc_arg));
273 arg->argc++;
274 arg->argv = mandoc_realloc
275 (arg->argv, arg->argc * sizeof(struct mdoc_argv));
277 (void)memcpy(&arg->argv[(int)arg->argc - 1],
278 &tmp, sizeof(struct mdoc_argv));
280 return(ARGV_ARG);
284 void
285 mdoc_argv_free(struct mdoc_arg *p)
287 int i, j;
289 if (NULL == p)
290 return;
292 if (p->refcnt) {
293 --(p->refcnt);
294 if (p->refcnt)
295 return;
297 assert(p->argc);
299 /* LINTED */
300 for (i = 0; i < (int)p->argc; i++) {
301 if (0 == p->argv[i].sz)
302 continue;
303 if (NULL == p->argv[i].value)
304 continue;
306 /* LINTED */
307 for (j = 0; j < (int)p->argv[i].sz; j++)
308 if (p->argv[i].value[j])
309 free(p->argv[i].value[j]);
311 free(p->argv[i].value);
314 free(p->argv);
315 free(p);
320 mdoc_zargs(struct mdoc *m, int line, int *pos,
321 char *buf, int flags, char **v)
324 return(args(m, line, pos, buf, flags, v));
329 mdoc_args(struct mdoc *m, int line,
330 int *pos, char *buf, int tok, char **v)
332 int fl, c, i;
333 struct mdoc_node *n;
335 fl = (0 == tok) ? 0 : mdoc_argflags[tok];
337 if (MDOC_It != tok)
338 return(args(m, line, pos, buf, fl, v));
341 * The `It' macro is a special case, as it acquires parameters from its
342 * parent `Bl' context, specifically, we're concerned with -column.
345 for (n = m->last; n; n = n->parent)
346 if (MDOC_BLOCK == n->type && MDOC_Bl == n->tok)
347 break;
349 assert(n);
350 c = (int)(n->args ? n->args->argc : 0);
351 assert(c > 0);
353 /* LINTED */
354 for (i = 0; i < c; i++) {
355 if (MDOC_Column != n->args->argv[i].arg)
356 continue;
357 fl |= ARGS_TABSEP;
358 fl &= ~ARGS_DELIM;
359 break;
362 return(args(m, line, pos, buf, fl, v));
366 static int
367 args(struct mdoc *m, int line, int *pos,
368 char *buf, int fl, char **v)
370 int i;
371 char *p, *pp;
374 * Parse out the terms (like `val' in `.Xx -arg val' or simply
375 * `.Xx val'), which can have all sorts of properties:
377 * ARGS_DELIM: use special handling if encountering trailing
378 * delimiters in the form of [[::delim::][ ]+]+.
380 * ARGS_NOWARN: don't post warnings. This is only used when
381 * re-parsing delimiters, as the warnings have already been
382 * posted.
384 * ARGS_TABSEP: use special handling for tab/`Ta' separated
385 * phrases like in `Bl -column'.
388 assert(*pos);
389 assert(' ' != buf[*pos]);
391 if (0 == buf[*pos])
392 return(ARGS_EOLN);
395 * If the first character is a delimiter and we're to look for
396 * delimited strings, then pass down the buffer seeing if it
397 * follows the pattern of [[::delim::][ ]+]+.
400 if ((fl & ARGS_DELIM) && mdoc_iscdelim(buf[*pos])) {
401 for (i = *pos; buf[i]; ) {
402 if ( ! mdoc_iscdelim(buf[i]))
403 break;
404 i++;
405 if (0 == buf[i] || ' ' != buf[i])
406 break;
407 i++;
408 while (buf[i] && ' ' == buf[i])
409 i++;
412 if (0 == buf[i]) {
413 *v = &buf[*pos];
414 if (' ' != buf[i - 1])
415 return(ARGS_PUNCT);
416 if (ARGS_NOWARN & fl)
417 return(ARGS_PUNCT);
418 if ( ! mdoc_pwarn(m, line, *pos, ETAILWS))
419 return(ARGS_ERROR);
420 return(ARGS_PUNCT);
424 *v = &buf[*pos];
427 * First handle TABSEP items, restricted to `Bl -column'. This
428 * ignores conventional token parsing and instead uses tabs or
429 * `Ta' macros to separate phrases. Phrases are parsed again
430 * for arguments at a later phase.
433 if (ARGS_TABSEP & fl) {
434 /* Scan ahead to tab (can't be escaped). */
435 p = strchr(*v, '\t');
437 /* Scan ahead to unescaped `Ta'. */
438 for (pp = *v; ; pp++) {
439 if (NULL == (pp = strstr(pp, "Ta")))
440 break;
441 if (pp > *v && ' ' != *(pp - 1))
442 continue;
443 if (' ' == *(pp + 2) || 0 == *(pp + 2))
444 break;
448 * Adjust new-buffer position to be beyond delimiter
449 * mark (e.g., Ta -> end + 2).
451 if (p && pp) {
452 *pos += pp < p ? 2 : 1;
453 p = pp < p ? pp : p;
454 } else if (p && ! pp) {
455 *pos += 1;
456 } else if (pp && ! p) {
457 p = pp;
458 *pos += 2;
459 } else
460 p = strchr(*v, 0);
462 /* Whitespace check for eoln case... */
463 if (0 == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl))
464 if ( ! mdoc_pwarn(m, line, *pos, ETAILWS))
465 return(ARGS_ERROR);
467 *pos += (int)(p - *v);
469 /* Strip delimiter's preceding whitespace. */
470 pp = p - 1;
471 while (pp > *v && ' ' == *pp) {
472 if (pp > *v && '\\' == *(pp - 1))
473 break;
474 pp--;
476 *(pp + 1) = 0;
478 /* Strip delimiter's proceeding whitespace. */
479 for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
480 /* Skip ahead. */ ;
482 return(ARGS_PHRASE);
486 * Process a quoted literal. A quote begins with a double-quote
487 * and ends with a double-quote NOT preceded by a double-quote.
488 * Whitespace is NOT involved in literal termination.
491 if ('\"' == buf[*pos]) {
492 *v = &buf[++(*pos)];
494 for ( ; buf[*pos]; (*pos)++) {
495 if ('\"' != buf[*pos])
496 continue;
497 if ('\"' != buf[*pos + 1])
498 break;
499 (*pos)++;
502 if (0 == buf[*pos]) {
503 if (ARGS_NOWARN & fl)
504 return(ARGS_QWORD);
505 if ( ! mdoc_pwarn(m, line, *pos, EQUOTTERM))
506 return(ARGS_ERROR);
507 return(ARGS_QWORD);
510 buf[(*pos)++] = 0;
512 if (0 == buf[*pos])
513 return(ARGS_QWORD);
515 while (' ' == buf[*pos])
516 (*pos)++;
518 if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
519 if ( ! mdoc_pwarn(m, line, *pos, ETAILWS))
520 return(ARGS_ERROR);
522 return(ARGS_QWORD);
526 * A non-quoted term progresses until either the end of line or
527 * a non-escaped whitespace.
530 for ( ; buf[*pos]; (*pos)++)
531 if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
532 break;
534 if (0 == buf[*pos])
535 return(ARGS_WORD);
537 buf[(*pos)++] = 0;
539 while (' ' == buf[*pos])
540 (*pos)++;
542 if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
543 if ( ! mdoc_pwarn(m, line, *pos, ETAILWS))
544 return(ARGS_ERROR);
546 return(ARGS_WORD);
550 static int
551 argv_a2arg(int tok, const char *p)
555 * Parse an argument identifier from its text. XXX - this
556 * should really be table-driven to clarify the code.
558 * If you add an argument to the list, make sure that you
559 * register it here with its one or more macros!
562 switch (tok) {
563 case (MDOC_An):
564 if (0 == strcmp(p, "split"))
565 return(MDOC_Split);
566 else if (0 == strcmp(p, "nosplit"))
567 return(MDOC_Nosplit);
568 break;
570 case (MDOC_Bd):
571 if (0 == strcmp(p, "ragged"))
572 return(MDOC_Ragged);
573 else if (0 == strcmp(p, "unfilled"))
574 return(MDOC_Unfilled);
575 else if (0 == strcmp(p, "filled"))
576 return(MDOC_Filled);
577 else if (0 == strcmp(p, "literal"))
578 return(MDOC_Literal);
579 else if (0 == strcmp(p, "file"))
580 return(MDOC_File);
581 else if (0 == strcmp(p, "offset"))
582 return(MDOC_Offset);
583 else if (0 == strcmp(p, "compact"))
584 return(MDOC_Compact);
585 else if (0 == strcmp(p, "centered"))
586 return(MDOC_Centred);
587 break;
589 case (MDOC_Bf):
590 if (0 == strcmp(p, "emphasis"))
591 return(MDOC_Emphasis);
592 else if (0 == strcmp(p, "literal"))
593 return(MDOC_Literal);
594 else if (0 == strcmp(p, "symbolic"))
595 return(MDOC_Symbolic);
596 break;
598 case (MDOC_Bk):
599 if (0 == strcmp(p, "words"))
600 return(MDOC_Words);
601 break;
603 case (MDOC_Bl):
604 if (0 == strcmp(p, "bullet"))
605 return(MDOC_Bullet);
606 else if (0 == strcmp(p, "dash"))
607 return(MDOC_Dash);
608 else if (0 == strcmp(p, "hyphen"))
609 return(MDOC_Hyphen);
610 else if (0 == strcmp(p, "item"))
611 return(MDOC_Item);
612 else if (0 == strcmp(p, "enum"))
613 return(MDOC_Enum);
614 else if (0 == strcmp(p, "tag"))
615 return(MDOC_Tag);
616 else if (0 == strcmp(p, "diag"))
617 return(MDOC_Diag);
618 else if (0 == strcmp(p, "hang"))
619 return(MDOC_Hang);
620 else if (0 == strcmp(p, "ohang"))
621 return(MDOC_Ohang);
622 else if (0 == strcmp(p, "inset"))
623 return(MDOC_Inset);
624 else if (0 == strcmp(p, "column"))
625 return(MDOC_Column);
626 else if (0 == strcmp(p, "width"))
627 return(MDOC_Width);
628 else if (0 == strcmp(p, "offset"))
629 return(MDOC_Offset);
630 else if (0 == strcmp(p, "compact"))
631 return(MDOC_Compact);
632 else if (0 == strcmp(p, "nested"))
633 return(MDOC_Nested);
634 break;
636 case (MDOC_Rv):
637 /* FALLTHROUGH */
638 case (MDOC_Ex):
639 if (0 == strcmp(p, "std"))
640 return(MDOC_Std);
641 break;
642 default:
643 break;
646 return(MDOC_ARG_MAX);
650 static int
651 argv_multi(struct mdoc *m, int line,
652 struct mdoc_argv *v, int *pos, char *buf)
654 int c;
655 char *p;
657 for (v->sz = 0; ; v->sz++) {
658 if ('-' == buf[*pos])
659 break;
660 c = args(m, line, pos, buf, 0, &p);
661 if (ARGS_ERROR == c)
662 return(0);
663 else if (ARGS_EOLN == c)
664 break;
666 if (0 == v->sz % MULTI_STEP)
667 v->value = mandoc_realloc(v->value,
668 (v->sz + MULTI_STEP) * sizeof(char *));
670 v->value[(int)v->sz] = mandoc_strdup(p);
673 return(1);
677 static int
678 argv_opt_single(struct mdoc *m, int line,
679 struct mdoc_argv *v, int *pos, char *buf)
681 int c;
682 char *p;
684 if ('-' == buf[*pos])
685 return(1);
687 c = args(m, line, pos, buf, 0, &p);
688 if (ARGS_ERROR == c)
689 return(0);
690 if (ARGS_EOLN == c)
691 return(1);
693 v->sz = 1;
694 v->value = mandoc_malloc(sizeof(char *));
695 v->value[0] = mandoc_strdup(p);
697 return(1);
702 * Parse a single, mandatory value from the stream.
704 static int
705 argv_single(struct mdoc *m, int line,
706 struct mdoc_argv *v, int *pos, char *buf)
708 int c, ppos;
709 char *p;
711 ppos = *pos;
713 c = args(m, line, pos, buf, 0, &p);
714 if (ARGS_ERROR == c)
715 return(0);
716 if (ARGS_EOLN == c)
717 return(mdoc_perr(m, line, ppos, EARGVAL));
719 v->sz = 1;
720 v->value = mandoc_malloc(sizeof(char *));
721 v->value[0] = mandoc_strdup(p);
723 return(1);
728 * Determine rules for parsing arguments. Arguments can either accept
729 * no parameters, an optional single parameter, one parameter, or
730 * multiple parameters.
732 static int
733 argv(struct mdoc *mdoc, int line,
734 struct mdoc_argv *v, int *pos, char *buf)
737 v->sz = 0;
738 v->value = NULL;
740 switch (mdoc_argvflags[v->arg]) {
741 case (ARGV_SINGLE):
742 return(argv_single(mdoc, line, v, pos, buf));
743 case (ARGV_MULTI):
744 return(argv_multi(mdoc, line, v, pos, buf));
745 case (ARGV_OPT_SINGLE):
746 return(argv_opt_single(mdoc, line, v, pos, buf));
747 default:
748 /* ARGV_NONE */
749 break;
752 return(1);