-/* $Id: argv.c,v 1.26 2009/01/22 14:56:21 kristaps Exp $ */
+/* $Id: argv.c,v 1.38 2009/03/06 14:13:47 kristaps Exp $ */
/*
* Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
*
#include "private.h"
/*
- * Parse arguments and parameters of macros. Arguments follow the
- * syntax of `-arg [val [valN...]]', while parameters are free-form text
- * following arguments (if any). This file must correctly handle the
- * strange punctuation rules dictated by groff.
+ * Routines to parse arguments of macros. Arguments follow the syntax
+ * of `-arg [val [valN...]]'. Arguments come in all types: quoted
+ * arguments, multiple arguments per value, no-value arguments, etc.
*/
#define ARGS_QUOTED (1 << 0)
#define ARGS_DELIM (1 << 1)
#define ARGS_TABSEP (1 << 2)
+#define ARGS_ARGVLIKE (1 << 3)
-static int lookup(int, const char *);
+#define ARGV_NONE (1 << 0)
+#define ARGV_SINGLE (1 << 1)
+#define ARGV_MULTI (1 << 2)
+#define ARGV_OPT_SINGLE (1 << 3)
+
+static int argv_a2arg(int, const char *);
static int args(struct mdoc *, int, int *,
char *, int, char **);
-static int argv(struct mdoc *, int,
+static int argv(struct mdoc *, int, int,
struct mdoc_arg *, int *, char *);
static int argv_single(struct mdoc *, int,
struct mdoc_arg *, int *, char *);
+static int argv_opt_single(struct mdoc *, int,
+ struct mdoc_arg *, int *, char *);
static int argv_multi(struct mdoc *, int,
struct mdoc_arg *, int *, char *);
-static int postargv(struct mdoc *, int,
- const struct mdoc_arg *, int);
static int pwarn(struct mdoc *, int, int, int);
static int perr(struct mdoc *, int, int, int);
/* Error messages. */
#define EQUOTTERM (0)
-#define EOFFSET (1)
-#define EARGVAL (2)
-#define EARGMANY (3)
+#define EARGVAL (1)
+#define EARGMANY (2)
+
+/* Per-argument flags. */
+
+static int mdoc_argvflags[MDOC_ARG_MAX] = {
+ ARGV_NONE, /* MDOC_Split */
+ ARGV_NONE, /* MDOC_Nosplit */
+ ARGV_NONE, /* MDOC_Ragged */
+ ARGV_NONE, /* MDOC_Unfilled */
+ ARGV_NONE, /* MDOC_Literal */
+ ARGV_NONE, /* MDOC_File */
+ ARGV_SINGLE, /* MDOC_Offset */
+ ARGV_NONE, /* MDOC_Bullet */
+ ARGV_NONE, /* MDOC_Dash */
+ ARGV_NONE, /* MDOC_Hyphen */
+ ARGV_NONE, /* MDOC_Item */
+ ARGV_NONE, /* MDOC_Enum */
+ ARGV_NONE, /* MDOC_Tag */
+ ARGV_NONE, /* MDOC_Diag */
+ ARGV_NONE, /* MDOC_Hang */
+ ARGV_NONE, /* MDOC_Ohang */
+ ARGV_NONE, /* MDOC_Inset */
+ ARGV_MULTI, /* MDOC_Column */
+ ARGV_SINGLE, /* MDOC_Width */
+ ARGV_NONE, /* MDOC_Compact */
+ ARGV_SINGLE, /* MDOC_Std */
+ ARGV_NONE, /* MDOC_Filled */
+ ARGV_NONE, /* MDOC_Words */
+ ARGV_NONE, /* MDOC_Emphasis */
+ ARGV_NONE /* MDOC_Symbolic */
+};
static int mdoc_argflags[MDOC_MAX] = {
0, /* \" */
0, /* Ed */
0, /* Bl */
0, /* El */
- ARGS_DELIM, /* It */
+ 0, /* It */
ARGS_DELIM, /* Ad */
ARGS_DELIM, /* An */
ARGS_DELIM, /* Ar */
0, /* Ot */
ARGS_DELIM, /* Pa */
0, /* Rv */
- ARGS_DELIM, /* St */
+ ARGS_DELIM | ARGS_ARGVLIKE, /* St */
ARGS_DELIM, /* Va */
ARGS_DELIM, /* Vt */
ARGS_DELIM, /* Xr */
0, /* Hf */
0, /* Fr */
0, /* Ud */
+ 0, /* Lb */
};
c = mdoc_perr(mdoc, line, pos,
"unterminated quoted parameter");
break;
- case (EOFFSET):
- c = mdoc_perr(mdoc, line, pos,
- "invalid value for offset argument");
- break;
case (EARGVAL):
c = mdoc_perr(mdoc, line, pos,
"argument requires a value");
fl = (0 == tok) ? 0 : mdoc_argflags[tok];
/*
- * First see if we should use TABSEP (Bl -column). This
- * invalidates the use of ARGS_DELIM.
+ * Override per-macro argument flags with context-specific ones.
+ * As of now, this is only valid for `It' depending on its list
+ * context.
*/
- if (MDOC_It == tok) {
+ switch (tok) {
+ case (MDOC_It):
for (n = mdoc->last; n; n = n->parent)
- if (MDOC_BLOCK == n->type)
- if (MDOC_Bl == n->tok)
- break;
+ if (MDOC_BLOCK == n->type && MDOC_Bl == n->tok)
+ break;
+
assert(n);
c = (int)n->data.block.argc;
assert(c > 0);
+ /*
+ * Using `Bl -column' adds ARGS_TABSEP to the arguments
+ * and invalidates ARGS_DELIM. Using `Bl -diag' allows
+ * for quoted arguments.
+ */
+
/* LINTED */
for (i = 0; i < c; i++) {
- if (MDOC_Column != n->data.block.argv[i].arg)
- continue;
- fl |= ARGS_TABSEP;
- fl &= ~ARGS_DELIM;
- break;
+ switch (n->data.block.argv[i].arg) {
+ case (MDOC_Column):
+ fl |= ARGS_TABSEP;
+ fl &= ~ARGS_DELIM;
+ i = c;
+ break;
+ case (MDOC_Diag):
+ fl |= ARGS_QUOTED;
+ i = c;
+ break;
+ default:
+ break;
+ }
}
+ default:
+ break;
}
+ /* Continue parsing the arguments themselves... */
+
return(args(mdoc, line, pos, buf, fl, v));
}
args(struct mdoc *mdoc, int line,
int *pos, char *buf, int fl, char **v)
{
- int i, c;
+ int i;
char *p, *pp;
assert(*pos > 0);
if ( ! pwarn(mdoc, line, *pos, WQUOTPARM))
return(ARGS_ERROR);
- if ('-' == buf[*pos])
+ if ( ! (fl & ARGS_ARGVLIKE) && '-' == buf[*pos])
if ( ! pwarn(mdoc, line, *pos, WARGVPARM))
return(ARGS_ERROR);
*/
if ((fl & ARGS_DELIM) && mdoc_iscdelim(buf[*pos])) {
- for (i = *pos; (c = buf[i]); ) {
- if ( ! mdoc_iscdelim(c))
+ for (i = *pos; buf[i]; ) {
+ if ( ! mdoc_iscdelim(buf[i]))
break;
i++;
- if (0 == buf[i] || ! isspace(c))
+ /* There must be at least one space... */
+ if (0 == buf[i] || ! isspace((u_char)buf[i]))
break;
i++;
- while (buf[i] && isspace(c))
+ while (buf[i] && isspace((u_char)buf[i]))
i++;
}
if (0 == buf[i]) {
p++;
if (0 != *p)
*(p - 1) = 0;
- *pos += p - *v;
+ *pos += (int)(p - *v);
}
if (p && 0 == *p)
if (p > *v && ' ' == *(p - 1))
if ( ! pwarn(mdoc, line, *pos, WTAILWS))
return(0);
- *pos += p - *v;
+ *pos += (int)(p - *v);
return(ARGS_WORD);
}
/* Do non-tabsep look-ahead here. */
if ( ! (ARGS_TABSEP & fl))
- while ((c = buf[*pos])) {
- if (isspace(c))
+ while (buf[*pos]) {
+ if (isspace((u_char)buf[*pos]))
if ('\\' != buf[*pos - 1])
break;
(*pos)++;
return(ARGS_WORD);
if ( ! (ARGS_TABSEP & fl))
- while (buf[*pos] && isspace((int)buf[*pos]))
+ while (buf[*pos] && isspace((u_char)buf[*pos]))
(*pos)++;
if (buf[*pos])
if (0 == buf[*pos])
return(ARGS_QWORD);
- while (buf[*pos] && isspace((int)buf[*pos]))
+ while (buf[*pos] && isspace((u_char)buf[*pos]))
(*pos)++;
if (buf[*pos])
static int
-lookup(int tok, const char *argv)
+argv_a2arg(int tok, const char *argv)
{
+ /*
+ * Parse an argument identifier from its text. XXX - this
+ * should really be table-driven to clarify the code.
+ *
+ * If you add an argument to the list, make sure that you
+ * register it here with its one or more macros!
+ */
+
switch (tok) {
case (MDOC_An):
if (xstrcmp(argv, "split"))
if (xstrcmp(argv, "std"))
return(MDOC_Std);
break;
-
- case (MDOC_St):
- if (xstrcmp(argv, "p1003.1-88"))
- return(MDOC_p1003_1_88);
- else if (xstrcmp(argv, "p1003.1-90"))
- return(MDOC_p1003_1_90);
- else if (xstrcmp(argv, "p1003.1-96"))
- return(MDOC_p1003_1_96);
- else if (xstrcmp(argv, "p1003.1-2001"))
- return(MDOC_p1003_1_2001);
- else if (xstrcmp(argv, "p1003.1-2004"))
- return(MDOC_p1003_1_2004);
- else if (xstrcmp(argv, "p1003.1"))
- return(MDOC_p1003_1);
- else if (xstrcmp(argv, "p1003.1b"))
- return(MDOC_p1003_1b);
- else if (xstrcmp(argv, "p1003.1b-93"))
- return(MDOC_p1003_1b_93);
- else if (xstrcmp(argv, "p1003.1c-95"))
- return(MDOC_p1003_1c_95);
- else if (xstrcmp(argv, "p1003.1g-2000"))
- return(MDOC_p1003_1g_2000);
- else if (xstrcmp(argv, "p1003.2-92"))
- return(MDOC_p1003_2_92);
- else if (xstrcmp(argv, "p1003.2-95"))
- return(MDOC_p1387_2_95);
- else if (xstrcmp(argv, "p1003.2"))
- return(MDOC_p1003_2);
- else if (xstrcmp(argv, "p1387.2-95"))
- return(MDOC_p1387_2);
- else if (xstrcmp(argv, "isoC-90"))
- return(MDOC_isoC_90);
- else if (xstrcmp(argv, "isoC-amd1"))
- return(MDOC_isoC_amd1);
- else if (xstrcmp(argv, "isoC-tcor1"))
- return(MDOC_isoC_tcor1);
- else if (xstrcmp(argv, "isoC-tcor2"))
- return(MDOC_isoC_tcor2);
- else if (xstrcmp(argv, "isoC-99"))
- return(MDOC_isoC_99);
- else if (xstrcmp(argv, "ansiC"))
- return(MDOC_ansiC);
- else if (xstrcmp(argv, "ansiC-89"))
- return(MDOC_ansiC_89);
- else if (xstrcmp(argv, "ansiC-99"))
- return(MDOC_ansiC_99);
- else if (xstrcmp(argv, "ieee754"))
- return(MDOC_ieee754);
- else if (xstrcmp(argv, "iso8802-3"))
- return(MDOC_iso8802_3);
- else if (xstrcmp(argv, "xpg3"))
- return(MDOC_xpg3);
- else if (xstrcmp(argv, "xpg4"))
- return(MDOC_xpg4);
- else if (xstrcmp(argv, "xpg4.2"))
- return(MDOC_xpg4_2);
- else if (xstrcmp(argv, "xpg4.3"))
- return(MDOC_xpg4_3);
- else if (xstrcmp(argv, "xbd5"))
- return(MDOC_xbd5);
- else if (xstrcmp(argv, "xcu5"))
- return(MDOC_xcu5);
- else if (xstrcmp(argv, "xsh5"))
- return(MDOC_xsh5);
- else if (xstrcmp(argv, "xns5"))
- return(MDOC_xns5);
- else if (xstrcmp(argv, "xns5.2d2.0"))
- return(MDOC_xns5_2d2_0);
- else if (xstrcmp(argv, "xcurses4.2"))
- return(MDOC_xcurses4_2);
- else if (xstrcmp(argv, "susv2"))
- return(MDOC_susv2);
- else if (xstrcmp(argv, "susv3"))
- return(MDOC_susv3);
- else if (xstrcmp(argv, "svid4"))
- return(MDOC_svid4);
- break;
-
default:
break;
}
}
-static int
-postargv(struct mdoc *mdoc, int line, const struct mdoc_arg *v, int pos)
-{
-
- switch (v->arg) {
- case (MDOC_Offset):
- assert(v->value);
- assert(v->value[0]);
- if (xstrcmp(v->value[0], "left"))
- break;
- if (xstrcmp(v->value[0], "right"))
- break;
- if (xstrcmp(v->value[0], "center"))
- break;
- if (xstrcmp(v->value[0], "indent"))
- break;
- if (xstrcmp(v->value[0], "indent-two"))
- break;
- return(perr(mdoc, line, pos, EOFFSET));
- default:
- break;
- }
-
- return(1);
-}
-
-
static int
argv_multi(struct mdoc *mdoc, int line,
struct mdoc_arg *v, int *pos, char *buf)
return(0);
} else if (ARGS_EOLN == c)
break;
- v->value[v->sz] = p;
+ v->value[(int)v->sz] = p;
}
if (0 < v->sz && v->sz < MDOC_LINEARG_MAX)
}
+static int
+argv_opt_single(struct mdoc *mdoc, int line,
+ struct mdoc_arg *v, int *pos, char *buf)
+{
+ int c, ppos;
+ char *p;
+
+ ppos = *pos;
+
+ if ('-' == buf[*pos])
+ return(1);
+
+ c = args(mdoc, line, pos, buf, ARGS_QUOTED, &p);
+ if (ARGS_ERROR == c)
+ return(0);
+ if (ARGS_EOLN == c)
+ return(1);
+
+ v->sz = 1;
+ v->value = xcalloc(1, sizeof(char *));
+ v->value[0] = p;
+ return(1);
+}
+
+
+/*
+ * Parse a single, mandatory value from the stream.
+ */
static int
argv_single(struct mdoc *mdoc, int line,
struct mdoc_arg *v, int *pos, char *buf)
if (ARGS_ERROR == c)
return(0);
if (ARGS_EOLN == c)
- return(perr(mdoc, line, ppos, EARGVAL));
+ return(perr(mdoc, line, ppos, EARGVAL));
v->sz = 1;
v->value = xcalloc(1, sizeof(char *));
}
+/*
+ * Determine rules for parsing arguments. Arguments can either accept
+ * no parameters, an optional single parameter, one parameter, or
+ * multiple parameters.
+ */
static int
-argv(struct mdoc *mdoc, int line,
+argv(struct mdoc *mdoc, int tok, int line,
struct mdoc_arg *v, int *pos, char *buf)
{
+ int fl;
v->sz = 0;
v->value = NULL;
+ fl = mdoc_argvflags[v->arg];
- switch (v->arg) {
- case(MDOC_Std):
- /* FALLTHROUGH */
- case(MDOC_Width):
- /* FALLTHROUGH */
- case(MDOC_Offset):
+ /*
+ * Override the default per-argument value.
+ */
+
+ switch (tok) {
+ case (MDOC_Ex):
+ fl = ARGV_OPT_SINGLE;
+ break;
+ default:
+ break;
+ }
+
+ switch (fl) {
+ case (ARGV_SINGLE):
return(argv_single(mdoc, line, v, pos, buf));
- case(MDOC_Column):
+ case (ARGV_MULTI):
return(argv_multi(mdoc, line, v, pos, buf));
+ case (ARGV_OPT_SINGLE):
+ return(argv_opt_single(mdoc, line, v, pos, buf));
default:
+ /* ARGV_NONE */
break;
}
}
+/*
+ * Parse an argument from line text. This comes in the form of -key
+ * [value0...], which may either have a single mandatory value, at least
+ * one mandatory value, an optional single value, or no value.
+ */
int
mdoc_argv(struct mdoc *mdoc, int line, int tok,
struct mdoc_arg *v, int *pos, char *buf)
{
- int i, ppos;
+ int i;
char *p;
(void)memset(v, 0, sizeof(struct mdoc_arg));
if (0 == buf[*pos])
return(ARGV_EOLN);
- assert( ! isspace((int)buf[*pos]));
+ assert( ! isspace((u_char)buf[*pos]));
if ('-' != buf[*pos])
return(ARGV_WORD);
/* LINTED */
while (buf[*pos]) {
- if (isspace((int)buf[*pos]))
+ if (isspace((u_char)buf[*pos]))
if ('\\' != buf[*pos - 1])
break;
(*pos)++;
if (buf[*pos])
buf[(*pos)++] = 0;
- if (MDOC_ARG_MAX == (v->arg = lookup(tok, p))) {
+ /*
+ * We now parse out the per-macro arguments. XXX - this can be
+ * made much cleaner using per-argument tables. See argv_a2arg
+ * for details.
+ */
+
+ if (MDOC_ARG_MAX == (v->arg = argv_a2arg(tok, p))) {
if ( ! pwarn(mdoc, line, i, WARGVPARM))
return(ARGV_ERROR);
return(ARGV_WORD);
}
- while (buf[*pos] && isspace((int)buf[*pos]))
+ while (buf[*pos] && isspace((u_char)buf[*pos]))
(*pos)++;
/* FIXME: whitespace if no value. */
- ppos = *pos;
- if ( ! argv(mdoc, line, v, pos, buf))
- return(ARGV_ERROR);
- if ( ! postargv(mdoc, line, v, ppos))
+ if ( ! argv(mdoc, tok, line, v, pos, buf))
return(ARGV_ERROR);
return(ARGV_ARG);
git.ckatri.com / mandoc.git / blobdiff