1 /* $Id: mdoc_argv.c,v 1.18 2009/10/27 21:40:07 schwarze Exp $ */
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>
28 * Routines to parse arguments of macros. Arguments follow the syntax
29 * of `-arg [val [valN...]]'. Arguments come in all types: quoted
30 * arguments, multiple arguments per value, no-value arguments, etc.
32 * There's no limit to the number or arguments that may be allocated.
35 #define ARGV_NONE (1 << 0)
36 #define ARGV_SINGLE (1 << 1)
37 #define ARGV_MULTI (1 << 2)
38 #define ARGV_OPT_SINGLE (1 << 3)
42 static int argv_a2arg(int, const char *);
43 static int args(struct mdoc *, int, int *,
44 char *, int, char **);
45 static int argv(struct mdoc *, int,
46 struct mdoc_argv *, int *, char *);
47 static int argv_single(struct mdoc *, int,
48 struct mdoc_argv *, int *, char *);
49 static int argv_opt_single(struct mdoc *, int,
50 struct mdoc_argv *, int *, char *);
51 static int argv_multi(struct mdoc *, int,
52 struct mdoc_argv *, int *, char *);
54 /* Per-argument flags. */
56 static int mdoc_argvflags[MDOC_ARG_MAX] = {
57 ARGV_NONE, /* MDOC_Split */
58 ARGV_NONE, /* MDOC_Nosplit */
59 ARGV_NONE, /* MDOC_Ragged */
60 ARGV_NONE, /* MDOC_Unfilled */
61 ARGV_NONE, /* MDOC_Literal */
62 ARGV_SINGLE, /* MDOC_File */
63 ARGV_OPT_SINGLE, /* MDOC_Offset */
64 ARGV_NONE, /* MDOC_Bullet */
65 ARGV_NONE, /* MDOC_Dash */
66 ARGV_NONE, /* MDOC_Hyphen */
67 ARGV_NONE, /* MDOC_Item */
68 ARGV_NONE, /* MDOC_Enum */
69 ARGV_NONE, /* MDOC_Tag */
70 ARGV_NONE, /* MDOC_Diag */
71 ARGV_NONE, /* MDOC_Hang */
72 ARGV_NONE, /* MDOC_Ohang */
73 ARGV_NONE, /* MDOC_Inset */
74 ARGV_MULTI, /* MDOC_Column */
75 ARGV_SINGLE, /* MDOC_Width */
76 ARGV_NONE, /* MDOC_Compact */
77 ARGV_NONE, /* MDOC_Std */
78 ARGV_NONE, /* MDOC_Filled */
79 ARGV_NONE, /* MDOC_Words */
80 ARGV_NONE, /* MDOC_Emphasis */
81 ARGV_NONE, /* MDOC_Symbolic */
82 ARGV_NONE /* MDOC_Symbolic */
85 static int mdoc_argflags[MDOC_MAX] = {
146 ARGS_DELIM, /* Bsx */
196 ARGS_DELIM, /* Brq */
198 ARGS_DELIM, /* Brc */
211 * Parse an argument from line text. This comes in the form of -key
212 * [value0...], which may either have a single mandatory value, at least
213 * one mandatory value, an optional single value, or no value.
216 mdoc_argv(struct mdoc *m, int line, int tok,
217 struct mdoc_arg **v, int *pos, char *buf)
220 struct mdoc_argv tmp;
221 struct mdoc_arg *arg;
226 assert(' ' != buf[*pos]);
228 /* Parse through to the first unescaped space. */
236 if (' ' == buf[*pos])
237 if ('\\' != buf[*pos - 1])
242 /* XXX - save zeroed byte, if not an argument. */
250 (void)memset(&tmp, 0, sizeof(struct mdoc_argv));
254 /* See if our token accepts the argument. */
256 if (MDOC_ARG_MAX == (tmp.arg = argv_a2arg(tok, p))) {
257 /* XXX - restore saved zeroed byte. */
263 while (buf[*pos] && ' ' == buf[*pos])
266 if ( ! argv(m, line, &tmp, pos, buf))
269 if (NULL == (arg = *v)) {
270 *v = calloc(1, sizeof(struct mdoc_arg));
272 (void)mdoc_nerr(m, m->last, EMALLOC);
279 arg->argv = realloc(arg->argv, arg->argc *
280 sizeof(struct mdoc_argv));
282 if (NULL == arg->argv) {
283 (void)mdoc_nerr(m, m->last, EMALLOC);
287 (void)memcpy(&arg->argv[(int)arg->argc - 1],
288 &tmp, sizeof(struct mdoc_argv));
295 mdoc_argv_free(struct mdoc_arg *p)
310 for (i = 0; i < (int)p->argc; i++) {
311 if (0 == p->argv[i].sz)
313 if (NULL == p->argv[i].value)
317 for (j = 0; j < (int)p->argv[i].sz; j++)
318 if (p->argv[i].value[j])
319 free(p->argv[i].value[j]);
321 free(p->argv[i].value);
330 mdoc_zargs(struct mdoc *m, int line, int *pos,
331 char *buf, int flags, char **v)
334 return(args(m, line, pos, buf, flags, v));
339 mdoc_args(struct mdoc *m, int line,
340 int *pos, char *buf, int tok, char **v)
345 fl = (0 == tok) ? 0 : mdoc_argflags[tok];
348 return(args(m, line, pos, buf, fl, v));
351 * The `It' macro is a special case, as it acquires parameters from its
352 * parent `Bl' context, specifically, we're concerned with -column.
355 for (n = m->last; n; n = n->parent)
356 if (MDOC_BLOCK == n->type && MDOC_Bl == n->tok)
360 c = (int)(n->args ? n->args->argc : 0);
364 for (i = 0; i < c; i++) {
365 if (MDOC_Column != n->args->argv[i].arg)
372 return(args(m, line, pos, buf, fl, v));
377 args(struct mdoc *m, int line, int *pos,
378 char *buf, int fl, char **v)
384 * Parse out the terms (like `val' in `.Xx -arg val' or simply
385 * `.Xx val'), which can have all sorts of properties:
387 * ARGS_DELIM: use special handling if encountering trailing
388 * delimiters in the form of [[::delim::][ ]+]+.
390 * ARGS_NOWARN: don't post warnings. This is only used when
391 * re-parsing delimiters, as the warnings have already been
394 * ARGS_TABSEP: use special handling for tab/`Ta' separated
395 * phrases like in `Bl -column'.
399 assert(' ' != buf[*pos]);
405 * If the first character is a delimiter and we're to look for
406 * delimited strings, then pass down the buffer seeing if it
407 * follows the pattern of [[::delim::][ ]+]+.
410 if ((fl & ARGS_DELIM) && mdoc_iscdelim(buf[*pos])) {
411 for (i = *pos; buf[i]; ) {
412 if ( ! mdoc_iscdelim(buf[i]))
415 if (0 == buf[i] || ' ' != buf[i])
418 while (buf[i] && ' ' == buf[i])
424 if (' ' != buf[i - 1])
426 if (ARGS_NOWARN & fl)
428 if ( ! mdoc_pwarn(m, line, *pos, ETAILWS))
437 * First handle TABSEP items, restricted to `Bl -column'. This
438 * ignores conventional token parsing and instead uses tabs or
439 * `Ta' macros to separate phrases. Phrases are parsed again
440 * for arguments at a later phase.
443 if (ARGS_TABSEP & fl) {
444 /* Scan ahead to tab (can't be escaped). */
445 p = strchr(*v, '\t');
447 /* Scan ahead to unescaped `Ta'. */
448 for (pp = *v; ; pp++) {
449 if (NULL == (pp = strstr(pp, "Ta")))
451 if (pp > *v && ' ' != *(pp - 1))
453 if (' ' == *(pp + 2) || 0 == *(pp + 2))
458 * Adjust new-buffer position to be beyond delimiter
459 * mark (e.g., Ta -> end + 2).
462 *pos += pp < p ? 2 : 1;
464 } else if (p && ! pp) {
466 } else if (pp && ! p) {
472 /* Whitespace check for eoln case... */
473 if (0 == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl))
474 if ( ! mdoc_pwarn(m, line, *pos, ETAILWS))
477 *pos += (int)(p - *v);
479 /* Strip delimiter's preceding whitespace. */
481 while (pp > *v && ' ' == *pp) {
482 if (pp > *v && '\\' == *(pp - 1))
488 /* Strip delimiter's proceeding whitespace. */
489 for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
496 * Process a quoted literal. A quote begins with a double-quote
497 * and ends with a double-quote NOT preceded by a double-quote.
498 * Whitespace is NOT involved in literal termination.
501 if ('\"' == buf[*pos]) {
504 for ( ; buf[*pos]; (*pos)++) {
505 if ('\"' != buf[*pos])
507 if ('\"' != buf[*pos + 1])
512 if (0 == buf[*pos]) {
513 if (ARGS_NOWARN & fl)
515 if ( ! mdoc_pwarn(m, line, *pos, EQUOTTERM))
525 while (' ' == buf[*pos])
528 if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
529 if ( ! mdoc_pwarn(m, line, *pos, ETAILWS))
536 * A non-quoted term progresses until either the end of line or
537 * a non-escaped whitespace.
540 for ( ; buf[*pos]; (*pos)++)
541 if (' ' == buf[*pos] && '\\' != buf[*pos - 1])
549 while (' ' == buf[*pos])
552 if (0 == buf[*pos] && ! (ARGS_NOWARN & fl))
553 if ( ! mdoc_pwarn(m, line, *pos, ETAILWS))
561 argv_a2arg(int tok, const char *p)
565 * Parse an argument identifier from its text. XXX - this
566 * should really be table-driven to clarify the code.
568 * If you add an argument to the list, make sure that you
569 * register it here with its one or more macros!
574 if (0 == strcmp(p, "split"))
576 else if (0 == strcmp(p, "nosplit"))
577 return(MDOC_Nosplit);
581 if (0 == strcmp(p, "ragged"))
583 else if (0 == strcmp(p, "unfilled"))
584 return(MDOC_Unfilled);
585 else if (0 == strcmp(p, "filled"))
587 else if (0 == strcmp(p, "literal"))
588 return(MDOC_Literal);
589 else if (0 == strcmp(p, "file"))
591 else if (0 == strcmp(p, "offset"))
593 else if (0 == strcmp(p, "compact"))
594 return(MDOC_Compact);
595 else if (0 == strcmp(p, "centered"))
596 return(MDOC_Centred);
600 if (0 == strcmp(p, "emphasis"))
601 return(MDOC_Emphasis);
602 else if (0 == strcmp(p, "literal"))
603 return(MDOC_Literal);
604 else if (0 == strcmp(p, "symbolic"))
605 return(MDOC_Symbolic);
609 if (0 == strcmp(p, "words"))
614 if (0 == strcmp(p, "bullet"))
616 else if (0 == strcmp(p, "dash"))
618 else if (0 == strcmp(p, "hyphen"))
620 else if (0 == strcmp(p, "item"))
622 else if (0 == strcmp(p, "enum"))
624 else if (0 == strcmp(p, "tag"))
626 else if (0 == strcmp(p, "diag"))
628 else if (0 == strcmp(p, "hang"))
630 else if (0 == strcmp(p, "ohang"))
632 else if (0 == strcmp(p, "inset"))
634 else if (0 == strcmp(p, "column"))
636 else if (0 == strcmp(p, "width"))
638 else if (0 == strcmp(p, "offset"))
640 else if (0 == strcmp(p, "compact"))
641 return(MDOC_Compact);
642 else if (0 == strcmp(p, "nested"))
649 if (0 == strcmp(p, "std"))
656 return(MDOC_ARG_MAX);
661 argv_multi(struct mdoc *m, int line,
662 struct mdoc_argv *v, int *pos, char *buf)
667 for (v->sz = 0; ; v->sz++) {
668 if ('-' == buf[*pos])
670 c = args(m, line, pos, buf, 0, &p);
673 else if (ARGS_EOLN == c)
676 if (0 == v->sz % MULTI_STEP) {
677 v->value = realloc(v->value,
678 (v->sz + MULTI_STEP) * sizeof(char *));
679 if (NULL == v->value) {
680 (void)mdoc_nerr(m, m->last, EMALLOC);
684 if (NULL == (v->value[(int)v->sz] = strdup(p)))
685 return(mdoc_nerr(m, m->last, EMALLOC));
693 argv_opt_single(struct mdoc *m, int line,
694 struct mdoc_argv *v, int *pos, char *buf)
699 if ('-' == buf[*pos])
702 c = args(m, line, pos, buf, 0, &p);
709 if (NULL == (v->value = calloc(1, sizeof(char *))))
710 return(mdoc_nerr(m, m->last, EMALLOC));
711 if (NULL == (v->value[0] = strdup(p)))
712 return(mdoc_nerr(m, m->last, EMALLOC));
719 * Parse a single, mandatory value from the stream.
722 argv_single(struct mdoc *m, int line,
723 struct mdoc_argv *v, int *pos, char *buf)
730 c = args(m, line, pos, buf, 0, &p);
734 return(mdoc_perr(m, line, ppos, EARGVAL));
737 if (NULL == (v->value = calloc(1, sizeof(char *))))
738 return(mdoc_nerr(m, m->last, EMALLOC));
739 if (NULL == (v->value[0] = strdup(p)))
740 return(mdoc_nerr(m, m->last, EMALLOC));
747 * Determine rules for parsing arguments. Arguments can either accept
748 * no parameters, an optional single parameter, one parameter, or
749 * multiple parameters.
752 argv(struct mdoc *mdoc, int line,
753 struct mdoc_argv *v, int *pos, char *buf)
759 switch (mdoc_argvflags[v->arg]) {
761 return(argv_single(mdoc, line, v, pos, buf));
763 return(argv_multi(mdoc, line, v, pos, buf));
764 case (ARGV_OPT_SINGLE):
765 return(argv_opt_single(mdoc, line, v, pos, buf));