===================================================================
RCS file: /cvs/mandoc/Attic/argv.c,v
retrieving revision 1.11
retrieving revision 1.21
diff -u -p -r1.11 -r1.21
--- mandoc/Attic/argv.c	2009/01/12 10:31:53	1.11
+++ mandoc/Attic/argv.c	2009/01/20 13:49:36	1.21
@@ -1,4 +1,4 @@
-/* $Id: argv.c,v 1.11 2009/01/12 10:31:53 kristaps Exp $ */
+/* $Id: argv.c,v 1.21 2009/01/20 13:49:36 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -25,10 +25,22 @@
 
 #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.
+ */
 
+/* FIXME: .It called with -column and quoted arguments. */
+
 static	int		 lookup(int, const char *);
-static	int		 parse(struct mdoc *, int, int,
+static	int		 parse(struct mdoc *, int,
 				struct mdoc_arg *, int *, char *);
+static	int		 parse_single(struct mdoc *, int, 
+				struct mdoc_arg *, int *, char *);
+static	int		 parse_multi(struct mdoc *, int, 
+				struct mdoc_arg *, int *, char *);
 static	int		 postparse(struct mdoc *, int, 
 				const struct mdoc_arg *, int);
 
@@ -42,22 +54,26 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char 
 		return(ARGS_EOLN);
 
 	if ('\"' == buf[*pos] && ! (fl & ARGS_QUOTED))
-		if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX_QUOTED))
+		if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX, "unexpected quoted parameter"))
 			return(ARGS_ERROR);
 
 	if ('-' == buf[*pos]) 
-		if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX_ARGLIKE))
+		if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX, "argument-like parameter"))
 			return(ARGS_ERROR);
 
 	if ((fl & ARGS_DELIM) && mdoc_iscdelim(buf[*pos])) {
+		/* 
+		 * If ARGS_DELIM, return ARGS_PUNCT if only space-separated
+		 * punctuation remains.  
+		 */
 		for (i = *pos; buf[i]; ) {
 			if ( ! mdoc_iscdelim(buf[i]))
 				break;
 			i++;
-			if (0 == buf[i] || ! isspace(buf[i]))
+			if (0 == buf[i] || ! isspace((int)buf[i]))
 				break;
 			i++;
-			while (buf[i] && isspace(buf[i]))
+			while (buf[i] && isspace((int)buf[i]))
 				i++;
 		}
 		if (0 == buf[i]) {
@@ -66,30 +82,54 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char 
 		}
 	}
 
-	/*
-	 * Parse routine for non-quoted string.  
-	 */
+	/* Parse routine for non-quoted string. */
 
-	if ('\"' != buf[*pos]) {
+	assert(*pos > 0);
+	if ('\"' != buf[*pos] || ! (ARGS_QUOTED & fl)) {
 		*v = &buf[*pos];
 
-		while (buf[*pos] && ! isspace(buf[*pos]))
-			(*pos)++;
+		/* FIXME: UGLY tab-sep processing. */
 
+		if (ARGS_TABSEP & fl)
+			while (buf[*pos]) {
+				if ('\t' == buf[*pos])
+					break;
+				if ('T' == buf[*pos]) {
+					(*pos)++;
+					if (0 == buf[*pos])
+						break;
+					if ('a' == buf[*pos]) {
+						buf[*pos - 1] = 0;
+						break;
+					}
+				}
+				(*pos)++;
+			}
+		else {
+			while (buf[*pos]) {
+				if (isspace((int)buf[*pos]))
+					if ('\\' != buf[*pos - 1])
+						break;
+				(*pos)++;
+			}
+		}
+
 		if (0 == buf[*pos])
 			return(ARGS_WORD);
 
 		buf[(*pos)++] = 0;
+
 		if (0 == buf[*pos])
 			return(ARGS_WORD);
 
-		while (buf[*pos] && isspace(buf[*pos]))
-			(*pos)++;
+		if ( ! (ARGS_TABSEP & fl))
+			while (buf[*pos] && isspace((int)buf[*pos]))
+				(*pos)++;
 
 		if (buf[*pos])
 			return(ARGS_WORD);
 
-		if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX_WS_EOLN))
+		if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_COMPAT, "whitespace at end-of-line"))
 			return(ARGS_ERROR);
 
 		return(ARGS_WORD);
@@ -107,24 +147,24 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char 
 		(*pos)++;
 
 	if (0 == buf[*pos]) {
-		(void)mdoc_perr(mdoc, line, *pos, ERR_SYNTAX_UNQUOTE);
+		(void)mdoc_perr(mdoc, line, *pos, "unterminated quoted parameter");
 		return(ARGS_ERROR);
 	}
 
 	buf[(*pos)++] = 0;
 	if (0 == buf[*pos])
-		return(ARGS_WORD);
+		return(ARGS_QWORD);
 
-	while (buf[*pos] && isspace(buf[*pos]))
+	while (buf[*pos] && isspace((int)buf[*pos]))
 		(*pos)++;
 
 	if (buf[*pos])
-		return(ARGS_WORD);
+		return(ARGS_QWORD);
 
-	if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX_WS_EOLN))
+	if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_COMPAT, "whitespace at end-of-line"))
 		return(ARGS_ERROR);
 
-	return(ARGS_WORD);
+	return(ARGS_QWORD);
 }
 
 
@@ -145,6 +185,8 @@ lookup(int tok, const char *argv)
 			return(MDOC_Ragged);
 		else if (xstrcmp(argv, "unfilled"))
 			return(MDOC_Unfilled);
+		else if (xstrcmp(argv, "filled"))
+			return(MDOC_Filled);
 		else if (xstrcmp(argv, "literal"))
 			return(MDOC_Literal);
 		else if (xstrcmp(argv, "file"))
@@ -308,7 +350,7 @@ postparse(struct mdoc *mdoc, int line, const struct md
 			break;
 		if (xstrcmp(v->value[0], "indent-two"))
 			break;
-		return(mdoc_perr(mdoc, line, pos, ERR_SYNTAX_ARGBAD));
+		return(mdoc_perr(mdoc, line, pos, "invalid offset value"));
 	default:
 		break;
 	}
@@ -318,65 +360,79 @@ postparse(struct mdoc *mdoc, int line, const struct md
 
 
 static int
-parse(struct mdoc *mdoc, int line, int tok, 
+parse_multi(struct mdoc *mdoc, int line, 
 		struct mdoc_arg *v, int *pos, char *buf)
 {
+	int		 c, ppos;
 	char		*p;
-	int		 c, ppos, i;
 
+	v->sz = 0;
+	v->value = xcalloc(MDOC_LINEARG_MAX, sizeof(char *));
+
 	ppos = *pos;
 
+	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);
+		if (ARGS_ERROR == c) {
+			free(v->value);
+			return(0);
+		} else if (ARGS_EOLN == c)
+			break;
+		v->value[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"));
+}
+
+
+static int
+parse_single(struct mdoc *mdoc, int line, 
+		struct mdoc_arg *v, int *pos, char *buf)
+{
+	int		 c, ppos;
+	char		*p;
+
+	ppos = *pos;
+
+	c = mdoc_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"));
+
+	v->sz = 1;
+	v->value = xcalloc(1, sizeof(char *));
+	v->value[0] = p;
+	return(1);
+}
+
+
+static int
+parse(struct mdoc *mdoc, int line, 
+		struct mdoc_arg *v, int *pos, char *buf)
+{
+
+	v->sz = 0;
+	v->value = NULL;
+
 	switch (v->arg) {
 	case(MDOC_Std):
 		/* FALLTHROUGH */
 	case(MDOC_Width):
 		/* FALLTHROUGH */
 	case(MDOC_Offset):
-		/*
-		 * This has a single value for an argument.
-		 */
-		c = mdoc_args(mdoc, line, pos, buf, ARGS_QUOTED, &p);
-		if (ARGS_ERROR == c)
-			return(0);
-		else if (ARGS_EOLN != c) {
-			v->sz = 1;
-			v->value = xcalloc(1, sizeof(char *));
-			v->value[0] = p;
-			break;
-		}
-		return(mdoc_perr(mdoc, line, ppos, ERR_SYNTAX_ARGVAL));
-
+		return(parse_single(mdoc, line, v, pos, buf));
 	case(MDOC_Column):
-		/*
-		 * This has several value for a single argument.  We
-		 * pre-allocate a pointer array and don't let it exceed
-		 * this size.
-		 */
-		v->sz = 0;
-		v->value = xcalloc(MDOC_LINEARG_MAX, sizeof(char *));
-		for (i = 0; i < MDOC_LINEARG_MAX; i++) {
-			c = mdoc_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[i] = p;
-		}
-		if (0 == i) {
-			free(v->value);
-			return(mdoc_perr(mdoc, line, ppos, 
-						ERR_SYNTAX_ARGVAL));
-		} else if (MDOC_LINEARG_MAX == i)
-			return(mdoc_perr(mdoc, line, ppos, 
-						ERR_SYNTAX_ARGMANY));
-
-		v->sz = i;
-		break;
-
+		return(parse_multi(mdoc, line, v, pos, buf));
 	default:
-		v->sz = 0;
-		v->value = NULL;
 		break;
 	}
 
@@ -396,7 +452,7 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok,
 	if (0 == buf[*pos])
 		return(ARGV_EOLN);
 
-	assert( ! isspace(buf[*pos]));
+	assert( ! isspace((int)buf[*pos]));
 
 	if ('-' != buf[*pos])
 		return(ARGV_WORD);
@@ -407,24 +463,30 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok,
 	v->line = line;
 	v->pos = *pos;
 
-	while (buf[*pos] && ! isspace(buf[*pos]))
+	assert(*pos > 0);
+	while (buf[*pos]) {
+		if (isspace((int)buf[*pos])) 
+			if ('\\' != buf[*pos - 1])
+				break;
 		(*pos)++;
+	}
 
 	if (buf[*pos])
 		buf[(*pos)++] = 0;
 
 	if (MDOC_ARG_MAX == (v->arg = lookup(tok, argv))) {
-		(void)mdoc_pwarn(mdoc, line, i, WARN_SYNTAX_ARGLIKE);
+		if ( ! mdoc_pwarn(mdoc, line, i, WARN_SYNTAX, "argument-like parameter"))
+			return(ARGV_ERROR);
 		return(ARGV_WORD);
 	}
 
-	while (buf[*pos] && isspace(buf[*pos]))
+	while (buf[*pos] && isspace((int)buf[*pos]))
 		(*pos)++;
 
 	/* FIXME: whitespace if no value. */
 
 	ppos = *pos;
-	if ( ! parse(mdoc, line, tok, v, pos, buf))
+	if ( ! parse(mdoc, line, v, pos, buf))
 		return(ARGV_ERROR);
 	if ( ! postparse(mdoc, line, v, ppos))
 		return(ARGV_ERROR);