-/* $Id: argv.c,v 1.24 2009/01/21 17:56:32 kristaps Exp $ */
+/* $Id: argv.c,v 1.35 2009/03/03 22:28:21 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)
-static int lookup(int, const char *);
-static int parse(struct mdoc *, int,
+static int argv_a2arg(int, const char *);
+static int args(struct mdoc *, int, int *,
+ char *, int, char **);
+static int argv(struct mdoc *, int,
struct mdoc_arg *, int *, char *);
-static int parse_single(struct mdoc *, int,
+static int argv_single(struct mdoc *, int,
struct mdoc_arg *, int *, char *);
-static int parse_multi(struct mdoc *, int,
+static int argv_multi(struct mdoc *, int,
struct mdoc_arg *, int *, char *);
-static int postparse(struct mdoc *, int,
- const struct mdoc_arg *, int);
static int pwarn(struct mdoc *, int, int, int);
+static int perr(struct mdoc *, int, int, int);
+
+/* Warning messages. */
#define WQUOTPARM (0)
#define WARGVPARM (1)
+#define WCOLEMPTY (2)
+#define WTAILWS (3)
+
+/* Error messages. */
+
+#define EQUOTTERM (0)
+#define EARGVAL (1)
+#define EARGMANY (2)
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 */
};
+static int
+perr(struct mdoc *mdoc, int line, int pos, int code)
+{
+ int c;
+
+ switch (code) {
+ case (EQUOTTERM):
+ c = mdoc_perr(mdoc, line, pos,
+ "unterminated quoted parameter");
+ break;
+ case (EARGVAL):
+ c = mdoc_perr(mdoc, line, pos,
+ "argument requires a value");
+ break;
+ case (EARGMANY):
+ c = mdoc_perr(mdoc, line, pos,
+ "too many values for argument");
+ break;
+ default:
+ abort();
+ /* NOTREACHED */
+ }
+ return(c);
+}
+
+
static int
pwarn(struct mdoc *mdoc, int line, int pos, int code)
{
c = mdoc_pwarn(mdoc, line, pos, WARN_SYNTAX,
"argument-like parameter");
break;
+ case (WCOLEMPTY):
+ c = mdoc_pwarn(mdoc, line, pos, WARN_SYNTAX,
+ "last list column is empty");
+ break;
+ case (WTAILWS):
+ c = mdoc_pwarn(mdoc, line, pos, WARN_COMPAT,
+ "trailing whitespace");
+ break;
default:
abort();
/* NOTREACHED */
mdoc_args(struct mdoc *mdoc, int line,
int *pos, char *buf, int tok, char **v)
{
- int i, c, fl;
- char *p, *pp;
+ int fl, c, i;
struct mdoc_node *n;
- assert(*pos > 0);
-
- if (0 == buf[*pos])
- return(ARGS_EOLN);
-
fl = (0 == tok) ? 0 : mdoc_argflags[tok];
- if ('\"' == buf[*pos] && ! (fl & ARGS_QUOTED))
- if ( ! pwarn(mdoc, line, *pos, WQUOTPARM))
- return(ARGS_ERROR);
-
- if ('-' == buf[*pos])
- if ( ! pwarn(mdoc, line, *pos, WARGVPARM))
- return(ARGS_ERROR);
-
/*
* First see if we should use TABSEP (Bl -column). This
* invalidates the use of ARGS_DELIM.
assert(n);
c = (int)n->data.block.argc;
assert(c > 0);
+
+ /* LINTED */
for (i = 0; i < c; i++) {
- if (MDOC_Column != n->data.block.argv[i].arg)
- continue;
- fl |= ARGS_TABSEP;
- fl &= ~ARGS_DELIM;
+ 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;
+ }
}
}
+ return(args(mdoc, line, pos, buf, fl, v));
+}
+
+
+static int
+args(struct mdoc *mdoc, int line,
+ int *pos, char *buf, int fl, char **v)
+{
+ int i;
+ char *p, *pp;
+
+ assert(*pos > 0);
+
+ if (0 == buf[*pos])
+ return(ARGS_EOLN);
+
+ if ('\"' == buf[*pos] && ! (fl & ARGS_QUOTED))
+ if ( ! pwarn(mdoc, line, *pos, WQUOTPARM))
+ return(ARGS_ERROR);
+
+ if ('-' == buf[*pos])
+ if ( ! pwarn(mdoc, line, *pos, WARGVPARM))
+ return(ARGS_ERROR);
+
/*
* If the first character is a delimiter and we're to look for
* delimited strings, then pass down the buffer seeing if it
*/
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((int)buf[i]))
break;
i++;
- while (buf[i] && isspace(c))
+ while (buf[i] && isspace((int)buf[i]))
i++;
}
if (0 == buf[i]) {
p++;
if (0 != *p)
*(p - 1) = 0;
- else if (0 == *p)
- if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX, "empty final token")) /* FIXME: verbiage */
- return(0);
- *pos += p - *v;
+ *pos += (int)(p - *v);
}
- /* Configure the eoln case, too. */
+ if (p && 0 == *p)
+ if ( ! pwarn(mdoc, line, *pos, WCOLEMPTY))
+ return(0);
+ if (p && 0 == *p && p > *v && ' ' == *(p - 1))
+ if ( ! pwarn(mdoc, line, *pos, WTAILWS))
+ return(0);
- if (NULL == p) {
- p = strchr(*v, 0);
- assert(p);
+ if (p)
+ return(ARGS_WORD);
- /*if (p > *v && ' ' == *(p - 1))
- Warn about whitespace. */
+ /* Configure the eoln case, too. */
- *pos += p - *v;
- }
+ p = strchr(*v, 0);
+ assert(p);
+
+ if (p > *v && ' ' == *(p - 1))
+ if ( ! pwarn(mdoc, line, *pos, WTAILWS))
+ return(0);
+ *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((int)buf[*pos]))
if ('\\' != buf[*pos - 1])
break;
(*pos)++;
if (buf[*pos])
return(ARGS_WORD);
- if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_COMPAT, "whitespace at end-of-line"))
+ if ( ! pwarn(mdoc, line, *pos, WTAILWS))
return(ARGS_ERROR);
return(ARGS_WORD);
(*pos)++;
if (0 == buf[*pos]) {
- (void)mdoc_perr(mdoc, line, *pos, "unterminated quoted parameter");
+ (void)perr(mdoc, line, *pos, EQUOTTERM);
return(ARGS_ERROR);
}
if (buf[*pos])
return(ARGS_QWORD);
- if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_COMPAT, "whitespace at end-of-line"))
+ if ( ! pwarn(mdoc, line, *pos, WTAILWS))
return(ARGS_ERROR);
return(ARGS_QWORD);
static int
-lookup(int tok, const char *argv)
+argv_a2arg(int tok, const char *argv)
{
switch (tok) {
static int
-postparse(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(mdoc_perr(mdoc, line, pos, "invalid offset value"));
- default:
- break;
- }
-
- return(1);
-}
-
-
-static int
-parse_multi(struct mdoc *mdoc, int line,
+argv_multi(struct mdoc *mdoc, int line,
struct mdoc_arg *v, int *pos, char *buf)
{
int c, ppos;
for (v->sz = 0; v->sz < MDOC_LINEARG_MAX; v->sz++) {
if ('-' == buf[*pos])
break;
- c = mdoc_args(mdoc, line, pos, buf, ARGS_QUOTED, &p);
+ c = args(mdoc, line, pos, buf, ARGS_QUOTED, &p);
if (ARGS_ERROR == c) {
free(v->value);
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)
return(1);
free(v->value);
- return(mdoc_perr(mdoc, line, ppos, 0 == v->sz ?
- "argument requires a value" :
- "too many values to argument"));
+ if (0 == v->sz)
+ return(perr(mdoc, line, ppos, EARGVAL));
+
+ return(perr(mdoc, line, ppos, EARGMANY));
}
static int
-parse_single(struct mdoc *mdoc, int line,
+argv_single(struct mdoc *mdoc, int line,
struct mdoc_arg *v, int *pos, char *buf)
{
int c, ppos;
ppos = *pos;
- c = mdoc_args(mdoc, line, pos, buf, ARGS_QUOTED, &p);
+ c = args(mdoc, line, pos, buf, ARGS_QUOTED, &p);
if (ARGS_ERROR == c)
return(0);
if (ARGS_EOLN == c)
- return(mdoc_perr(mdoc, line, ppos, "argument requires a value"));
+ return(perr(mdoc, line, ppos, EARGVAL));
v->sz = 1;
v->value = xcalloc(1, sizeof(char *));
static int
-parse(struct mdoc *mdoc, int line,
+argv(struct mdoc *mdoc, int line,
struct mdoc_arg *v, int *pos, char *buf)
{
case(MDOC_Width):
/* FALLTHROUGH */
case(MDOC_Offset):
- return(parse_single(mdoc, line, v, pos, buf));
+ return(argv_single(mdoc, line, v, pos, buf));
case(MDOC_Column):
- return(parse_multi(mdoc, line, v, pos, buf));
+ return(argv_multi(mdoc, line, v, pos, buf));
default:
break;
}
mdoc_argv(struct mdoc *mdoc, int line, int tok,
struct mdoc_arg *v, int *pos, char *buf)
{
- int i, ppos;
- char *argv;
+ int i;
+ char *p;
(void)memset(v, 0, sizeof(struct mdoc_arg));
return(ARGV_WORD);
i = *pos;
- argv = &buf[++(*pos)];
+ p = &buf[++(*pos)];
v->line = line;
v->pos = *pos;
assert(*pos > 0);
+
+ /* LINTED */
while (buf[*pos]) {
if (isspace((int)buf[*pos]))
if ('\\' != buf[*pos - 1])
if (buf[*pos])
buf[(*pos)++] = 0;
- if (MDOC_ARG_MAX == (v->arg = lookup(tok, argv))) {
- if ( ! mdoc_pwarn(mdoc, line, i, WARN_SYNTAX, "argument-like parameter"))
+ if (MDOC_ARG_MAX == (v->arg = argv_a2arg(tok, p))) {
+ if ( ! pwarn(mdoc, line, i, WARGVPARM))
return(ARGV_ERROR);
return(ARGV_WORD);
}
/* FIXME: whitespace if no value. */
- ppos = *pos;
- if ( ! parse(mdoc, line, v, pos, buf))
- return(ARGV_ERROR);
- if ( ! postparse(mdoc, line, v, ppos))
+ if ( ! argv(mdoc, line, v, pos, buf))
return(ARGV_ERROR);
return(ARGV_ARG);
git.ckatri.com / mandoc.git / blobdiff