===================================================================
RCS file: /cvs/mandoc/Attic/argv.c,v
retrieving revision 1.14
retrieving revision 1.24
diff -u -p -r1.14 -r1.24
--- mandoc/Attic/argv.c	2009/01/16 11:50:54	1.14
+++ mandoc/Attic/argv.c	2009/01/21 17:56:32	1.24
@@ -1,4 +1,4 @@
-/* $Id: argv.c,v 1.14 2009/01/16 11:50:54 kristaps Exp $ */
+/* $Id: argv.c,v 1.24 2009/01/21 17:56:32 kristaps Exp $ */
 /*
  * Copyright (c) 2008 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -25,9 +25,19 @@
 
 #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.
+ */
 
+#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, int,
+static	int		 parse(struct mdoc *, int,
 				struct mdoc_arg *, int *, char *);
 static	int		 parse_single(struct mdoc *, int, 
 				struct mdoc_arg *, int *, char *);
@@ -35,37 +45,202 @@ static	int		 parse_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);
 
+#define	WQUOTPARM	(0)
+#define	WARGVPARM	(1)
 
+static	int mdoc_argflags[MDOC_MAX] = {
+	0, /* \" */
+	0, /* Dd */
+	0, /* Dt */
+	0, /* Os */
+	0, /* Sh */
+	0, /* Ss */ 
+	ARGS_DELIM, /* Pp */ 
+	ARGS_DELIM, /* D1 */
+	ARGS_DELIM, /* Dl */
+	0, /* Bd */
+	0, /* Ed */
+	0, /* Bl */
+	0, /* El */
+	ARGS_DELIM, /* It */
+	ARGS_DELIM, /* Ad */ 
+	ARGS_DELIM, /* An */
+	ARGS_DELIM, /* Ar */
+	ARGS_QUOTED, /* Cd */
+	ARGS_DELIM, /* Cm */
+	ARGS_DELIM, /* Dv */ 
+	ARGS_DELIM, /* Er */ 
+	ARGS_DELIM, /* Ev */ 
+	0, /* Ex */
+	ARGS_DELIM | ARGS_QUOTED, /* Fa */ 
+	0, /* Fd */ 
+	ARGS_DELIM, /* Fl */
+	ARGS_DELIM | ARGS_QUOTED, /* Fn */ 
+	ARGS_DELIM | ARGS_QUOTED, /* Ft */ 
+	ARGS_DELIM, /* Ic */ 
+	0, /* In */ 
+	ARGS_DELIM, /* Li */
+	0, /* Nd */ 
+	ARGS_DELIM, /* Nm */ 
+	ARGS_DELIM, /* Op */
+	0, /* Ot */
+	ARGS_DELIM, /* Pa */
+	0, /* Rv */
+	ARGS_DELIM, /* St */ 
+	ARGS_DELIM, /* Va */
+	ARGS_DELIM, /* Vt */ 
+	ARGS_DELIM, /* Xr */
+	ARGS_QUOTED, /* %A */
+	ARGS_QUOTED, /* %B */
+	ARGS_QUOTED, /* %D */
+	ARGS_QUOTED, /* %I */
+	ARGS_QUOTED, /* %J */
+	ARGS_QUOTED, /* %N */
+	ARGS_QUOTED, /* %O */
+	ARGS_QUOTED, /* %P */
+	ARGS_QUOTED, /* %R */
+	ARGS_QUOTED, /* %T */
+	ARGS_QUOTED, /* %V */
+	ARGS_DELIM, /* Ac */
+	0, /* Ao */
+	ARGS_DELIM, /* Aq */
+	ARGS_DELIM, /* At */
+	ARGS_DELIM, /* Bc */
+	0, /* Bf */ 
+	0, /* Bo */
+	ARGS_DELIM, /* Bq */
+	ARGS_DELIM, /* Bsx */
+	ARGS_DELIM, /* Bx */
+	0, /* Db */
+	ARGS_DELIM, /* Dc */
+	0, /* Do */
+	ARGS_DELIM, /* Dq */
+	ARGS_DELIM, /* Ec */
+	0, /* Ef */
+	ARGS_DELIM, /* Em */ 
+	0, /* Eo */
+	ARGS_DELIM, /* Fx */
+	ARGS_DELIM, /* Ms */
+	ARGS_DELIM, /* No */
+	ARGS_DELIM, /* Ns */
+	ARGS_DELIM, /* Nx */
+	ARGS_DELIM, /* Ox */
+	ARGS_DELIM, /* Pc */
+	ARGS_DELIM, /* Pf */
+	0, /* Po */
+	ARGS_DELIM, /* Pq */
+	ARGS_DELIM, /* Qc */
+	ARGS_DELIM, /* Ql */
+	0, /* Qo */
+	ARGS_DELIM, /* Qq */
+	0, /* Re */
+	0, /* Rs */
+	ARGS_DELIM, /* Sc */
+	0, /* So */
+	ARGS_DELIM, /* Sq */
+	0, /* Sm */
+	ARGS_DELIM, /* Sx */
+	ARGS_DELIM, /* Sy */
+	ARGS_DELIM, /* Tn */
+	ARGS_DELIM, /* Ux */
+	ARGS_DELIM, /* Xc */
+	0, /* Xo */
+	0, /* Fo */ 
+	0, /* Fc */ 
+	0, /* Oo */
+	ARGS_DELIM, /* Oc */
+	0, /* Bk */
+	0, /* Ek */
+	0, /* Bt */
+	0, /* Hf */
+	0, /* Fr */
+	0, /* Ud */
+};
+
+
+static int
+pwarn(struct mdoc *mdoc, int line, int pos, int code)
+{
+	int		 c;
+
+	switch (code) {
+	case (WQUOTPARM):
+		c = mdoc_pwarn(mdoc, line, pos, WARN_SYNTAX, 
+				"unexpected quoted parameter");
+		break;
+	case (WARGVPARM):
+		c = mdoc_pwarn(mdoc, line, pos, WARN_SYNTAX, 
+				"argument-like parameter");
+		break;
+	default:
+		abort();
+		/* NOTREACHED */
+	}
+	return(c);
+}
+
+
 int
-mdoc_args(struct mdoc *mdoc, int line, int *pos, char *buf, int fl, char **v)
+mdoc_args(struct mdoc *mdoc, int line, 
+		int *pos, char *buf, int tok, char **v)
 {
-	int		 i;
+	int		  i, c, fl;
+	char		 *p, *pp;
+	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 ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX, "unexpected quoted parameter"))
+		if ( ! pwarn(mdoc, line, *pos, WQUOTPARM))
 			return(ARGS_ERROR);
 
 	if ('-' == buf[*pos]) 
-		if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX, "argument-like parameter"))
+		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.
+	 */
+
+	if (MDOC_It == tok) {
+		for (n = mdoc->last; n; n = n->parent)
+			if (MDOC_BLOCK == n->type)
+				if (MDOC_Bl == n->tok)
+					break;
+		assert(n);
+		c = (int)n->data.block.argc;
+		assert(c > 0);
+		for (i = 0; i < c; i++) {
+			if (MDOC_Column != n->data.block.argv[i].arg)
+				continue;
+			fl |= ARGS_TABSEP;
+			fl &= ~ARGS_DELIM;
+		}
+	}
+
+	/* 
+	 * If the first character is a delimiter and we're to look for
+	 * delimited strings, then pass down the buffer seeing if it
+	 * follows the pattern of [[::delim::][ ]+]+.
+	 */
+
 	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]))
+		for (i = *pos; (c = buf[i]); ) {
+			if ( ! mdoc_iscdelim(c))
 				break;
 			i++;
-			if (0 == buf[i] || ! isspace(buf[i]))
+			if (0 == buf[i] || ! isspace(c))
 				break;
 			i++;
-			while (buf[i] && isspace(buf[i]))
+			while (buf[i] && isspace(c))
 				i++;
 		}
 		if (0 == buf[i]) {
@@ -74,31 +249,105 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char 
 		}
 	}
 
-	/* Parse routine for non-quoted string. */
+	/* First parse non-quoted strings. */
 
-	if ('\"' != buf[*pos]) {
+	if ('\"' != buf[*pos] || ! (ARGS_QUOTED & fl)) {
 		*v = &buf[*pos];
 
-		/* FIXME: UGLY tab-sep processing. */
+		/* 
+		 * Thar be dragons here!  If we're tab-separated, search
+		 * ahead for either a tab or the `Ta' macro.  If a tab
+		 * is detected, it mustn't be escaped; if a `Ta' is
+		 * detected, it must be space-buffered before and after.
+		 * If either of these hold true, then prune out the
+		 * extra spaces and call it an argument.
+		 */
 
-		if (ARGS_TABSEP & fl)
-			while (buf[*pos]) {
-				if ('\t' == buf[*pos])
+		if (ARGS_TABSEP & fl) {
+			/* Scan ahead to unescaped tab. */
+
+			for (p = *v; ; p++) {
+				if (NULL == (p = strchr(p, '\t')))
 					break;
-				if ('T' == buf[*pos]) {
-					(*pos)++;
-					if (0 == buf[*pos])
+				if (p == *v)
+					break;
+				if ('\\' != *(p - 1))
+					break;
+			}
+
+			/* Scan ahead to unescaped `Ta'. */
+
+			for (pp = *v; ; pp++) {
+				if (NULL == (pp = strstr(pp, "Ta")))
+					break;
+				if (pp > *v && ' ' != *(pp - 1))
+					continue;
+				if (' ' == *(pp + 2) || 0 == *(pp + 2))
+					break;
+			}
+
+			/* Choose delimiter tab/Ta. */
+
+			if (p && pp)
+				p = (p < pp ? p : pp);
+			else if ( ! p && pp)
+				p = pp;
+
+			/* Strip delimiter's preceding whitespace. */
+
+			if (p && p > *v) {
+				pp = p - 1;
+				while (pp > *v && ' ' == *pp)
+					pp--;
+				if (pp == *v && ' ' == *pp) 
+					*pp = 0;
+				else if (' ' == *pp)
+					*(pp + 1) = 0;
+			}
+
+			/* ...in- and proceding whitespace. */
+
+			if (p && ('\t' != *p)) {
+				*p++ = 0;
+				*p++ = 0;
+			} else if (p)
+				*p++ = 0;
+
+			if (p) {
+				while (' ' == *p)
+					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;
+			} 
+
+			/* Configure the eoln case, too. */
+
+			if (NULL == p) {
+				p = strchr(*v, 0);
+				assert(p);
+
+				/*if (p > *v && ' ' == *(p - 1))
+					Warn about whitespace. */
+
+				*pos += p - *v;
+			}
+
+			return(ARGS_WORD);
+		} 
+
+		/* Do non-tabsep look-ahead here. */
+		
+		if ( ! (ARGS_TABSEP & fl))
+			while ((c = buf[*pos])) {
+				if (isspace(c))
+					if ('\\' != buf[*pos - 1])
 						break;
-					if ('a' == buf[*pos]) {
-						buf[*pos - 1] = 0;
-						break;
-					}
-				}
 				(*pos)++;
 			}
-		else
-			while (buf[*pos] && ! isspace(buf[*pos]))
-				(*pos)++;
 
 		if (0 == buf[*pos])
 			return(ARGS_WORD);
@@ -109,7 +358,7 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char 
 			return(ARGS_WORD);
 
 		if ( ! (ARGS_TABSEP & fl))
-			while (buf[*pos] && isspace(buf[*pos]))
+			while (buf[*pos] && isspace((int)buf[*pos]))
 				(*pos)++;
 
 		if (buf[*pos])
@@ -139,18 +388,18 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char 
 
 	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_COMPAT, "whitespace at end-of-line"))
 		return(ARGS_ERROR);
 
-	return(ARGS_WORD);
+	return(ARGS_QWORD);
 }
 
 
@@ -402,7 +651,7 @@ parse_single(struct mdoc *mdoc, int line, 
 
 
 static int
-parse(struct mdoc *mdoc, int line, int tok, 
+parse(struct mdoc *mdoc, int line, 
 		struct mdoc_arg *v, int *pos, char *buf)
 {
 
@@ -438,7 +687,7 @@ 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);
@@ -449,8 +698,13 @@ 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;
@@ -461,13 +715,13 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok,
 		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);