Fixed spacing (almost there).

4 files changed, 114 insertions, 43 deletions

diff --git a/mdocml.1 b/mdocml.1
index 7f507fdb..c5f86d66 100644
--- a/mdocml.1
+++ b/mdocml.1
@@ -53,10 +53,6 @@ The validate filter, specified by
 is the default filter.  It produces no output, so specifying
 .Fl o Ar outfile
 results in an empty file.
-.Qq Op Op Fl a ) , Fl b ,
-.Bf Em
-.Qq Op Op Fl a ) , Fl b ,
-.Ef
 .\" The following requests should be uncommented and used where appropriate.
 .\" This next request is for sections 2, 3, and 9 function return values only.
 .\" .Sh RETURN VALUES
@@ -78,7 +74,7 @@ results in an empty file.
 The
 .Nm
 utility was written by 
-.An Kristaps Dzonsons Aq kristaps@kth.se .
+.An Em Kristaps Dzonsons Aq kristaps@kth.se .
 .\"
 .Sh CAVEATS
 Most caveats of
@@ -99,6 +95,9 @@ All macro arguments may be quoted, instead of only some.
 .It 
 Blank lines raise warnings.
 .It 
-Non-ascii characters may cause issues.
+If terminating punctuation is found, then 
+.Em all
+remaining tokens are flushed after line scope is closed, not just the
+last one.
 .El
 .\" .Sh BUGS
diff --git a/private.h b/private.h
index 628f4c98..40d39270 100644
--- a/private.h
+++ b/private.h
@@ -214,7 +214,7 @@ struct	roffcb {
 	int	(*roffhead)(void *);
 	int	(*rofftail)(void *);
 	int	(*roffin)(void *, int, int *, char **);
-	int	(*roffdata)(void *, char *);
+	int	(*roffdata)(void *, int, char *);
 	int	(*roffout)(void *, int);
 	int	(*roffblkin)(void *, int, int *, char **);
 	int	(*roffblkout)(void *, int);
diff --git a/roff.c b/roff.c
index 6b77f35a..d692dad2 100644
--- a/roff.c
+++ b/roff.c
@@ -34,12 +34,10 @@
 
 /* FIXME: warn about empty lists. */
 
-/* FIXME: ; : } ) (etc.) after text macros? */
+/* FIXME: roff_layout and roff_text have identical-ish lower bodies. */
 
 /* FIXME: NAME section needs specific elements. */
 
-/* FIXME: don't print Os, just do roffbegin. */
-
 #define	ROFF_MAXARG	  32
 
 enum	roffd { 
@@ -70,8 +68,7 @@ struct	rofftok {
 	int		  flags;
 #define	ROFF_PARSED	 (1 << 0)		/* "Parsed". */
 #define	ROFF_CALLABLE	 (1 << 1)		/* "Callable". */
-#define	ROFF_QUOTES	 (1 << 2)		/* Quoted args. */
-#define	ROFF_SHALLOW	 (1 << 3)		/* Nesting block. */
+#define	ROFF_SHALLOW	 (1 << 2)		/* Nesting block. */
 };
 
 struct	roffarg {
@@ -179,9 +176,9 @@ static	const struct rofftok tokens[ROFF_MAX] = {
 	{roff_comment, NULL, NULL, NULL, 0, ROFF_COMMENT, 0 }, /* \" */
 	{     roff_Dd, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Dd */
 	{     roff_Dt, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Dt */
-	{     roff_Os, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_QUOTES }, /* Os */
-	{ roff_layout, NULL, NULL, NULL, ROFF_Sh, ROFF_LAYOUT, ROFF_PARSED }, /* Sh */
-	{ roff_layout, NULL, NULL, NULL, ROFF_Ss, ROFF_LAYOUT, ROFF_PARSED }, /* Ss */ 
+	{     roff_Os, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Os */
+	{ roff_layout, NULL, NULL, NULL, ROFF_Sh, ROFF_LAYOUT, 0 }, /* Sh */
+	{ roff_layout, NULL, NULL, NULL, ROFF_Ss, ROFF_LAYOUT, 0 }, /* Ss */ 
 	{   roff_text, NULL, NULL, NULL, ROFF_Pp, ROFF_TEXT, 0 }, /* Pp */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* D1 */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Dl */
@@ -189,11 +186,11 @@ static	const struct rofftok tokens[ROFF_MAX] = {
 	{  roff_close, NULL, NULL, NULL, ROFF_Bd, ROFF_LAYOUT, 0 }, /* Ed */
 	{ roff_layout, roffarg_Bl, NULL, roffchild_Bl, 0, ROFF_LAYOUT, 0 }, /* Bl */
 	{  roff_close, NULL, roffparent_El, NULL, ROFF_Bl, ROFF_LAYOUT, 0 }, /* El */
-	{ roff_layout, NULL, roffparent_It, NULL, ROFF_It, ROFF_LAYOUT, ROFF_SHALLOW }, /* It */
+	{ roff_layout, NULL, roffparent_It, NULL, ROFF_It, ROFF_LAYOUT, ROFF_PARSED | ROFF_SHALLOW }, /* It */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ad */ /* FIXME */
 	{   roff_text, roffarg_An, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* An */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ar */
-	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_QUOTES }, /* Cd */ /* XXX man.4 only */
+	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Cd */ /* XXX man.4 only */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Cm */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dv */ /* XXX needs arg */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Er */ /* XXX needs arg */
@@ -207,7 +204,7 @@ static	const struct rofftok tokens[ROFF_MAX] = {
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ic */ /* XXX needs arg */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* In */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Li */
-	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_QUOTES }, /* Nd */
+	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Nd */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Nm */ /* FIXME */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Op */
 	{   NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ot */ /* XXX deprecated */
@@ -441,7 +438,7 @@ static int
 textparse(const struct rofftree *tree, char *buf)
 {
 
-	return((*tree->cb.roffdata)(tree->arg, buf));
+	return((*tree->cb.roffdata)(tree->arg, 1, buf));
 }
 
 
@@ -696,7 +693,6 @@ rofffindtok(const char *buf)
 }
 
 
-#if notyet
 static int
 roffispunct(const char *p)
 {
@@ -737,7 +733,6 @@ roffispunct(const char *p)
 
 	return(0);
 }
-#endif
 
 
 static int
@@ -985,13 +980,16 @@ roff_layout(ROFFCALL_ARGS)
 		return(0);
 
 	if ( ! (ROFF_PARSED & tokens[tok].flags)) {
+		i = 0;
 		while (*argv) {
-			if ( ! (*tree->cb.roffdata)(tree->arg, *argv++))
+			if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++))
 				return(0);
+			i = 1;
 		}
 		return((*tree->cb.roffout)(tree->arg, tok));
 	}
 
+	i = 0;
 	while (*argv) {
 		if (ROFF_MAX != (c = rofffindcallable(*argv))) {
 			if (NULL == tokens[c].cb) {
@@ -1006,8 +1004,35 @@ roff_layout(ROFFCALL_ARGS)
 		}
 
 		assert(tree->arg);
-		if ( ! (*tree->cb.roffdata)(tree->arg, *argv++))
+		if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++))
 			return(0);
+		i = 1;
+	}
+
+	/* 
+	 * If we're the first parser (*argv == tree->cur) then purge out
+	 * any additional punctuation, should there be any remaining at
+	 * the end of line. 
+	 */
+
+	if (ROFF_PARSED & tokens[tok].flags && *argv) {
+		i = 0;
+		while (argv[i])
+			i++;
+
+		assert(i > 0);
+		if ( ! roffispunct(argv[--i]))
+			return(1);
+
+		while (i >= 0 && roffispunct(argv[i]))
+			i--;
+
+		assert(0 != i);
+		i++;
+
+		while (argv[i])
+			if ( ! (*tree->cb.roffdata)(tree->arg, 0, argv[i++]))
+				return(0);
 	}
 
 	return((*tree->cb.roffout)(tree->arg, tok));
@@ -1018,7 +1043,7 @@ roff_layout(ROFFCALL_ARGS)
 static int
 roff_text(ROFFCALL_ARGS) 
 {
-	int		 i, c, argcp[ROFF_MAXARG];
+	int		 i, j, first, c, argcp[ROFF_MAXARG];
 	char		*v, *argvp[ROFF_MAXARG];
 
 	if (ROFF_PRELUDE & tree->state) {
@@ -1027,6 +1052,9 @@ roff_text(ROFFCALL_ARGS)
 		return(0);
 	}
 
+	/* FIXME: breaks if passed from roff_layout. */
+	first = *argv == tree->cur;
+
 	i = 0;
 	argv++;
 
@@ -1047,17 +1075,35 @@ roff_text(ROFFCALL_ARGS)
 		return(0);
 
 	if ( ! (ROFF_PARSED & tokens[tok].flags)) {
+		i = 0;
 		while (*argv) {
-			if ( ! (*tree->cb.roffdata)(tree->arg, *argv++))
+			if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++))
 				return(0);
+			i = 1;
 		}
 		return((*tree->cb.roffout)(tree->arg, tok));
 	}
 
+	i = 0;
 	while (*argv) {
 		if (ROFF_MAX == (c = rofffindcallable(*argv))) {
-			if ( ! (*tree->cb.roffdata)(tree->arg, *argv++))
+			/* 
+			 * If all that remains is roff punctuation, then
+			 * close out our scope and return.
+			 */
+			if (roffispunct(*argv)) {
+				for (j = 0; argv[j]; j++)
+					if ( ! roffispunct(argv[j]))
+						break;
+				if (NULL == argv[j])
+					break;
+				i = 1;
+			}
+			
+			if ( ! (*tree->cb.roffdata)(tree->arg, i, *argv++))
 				return(0);
+
+			i = 1;
 			continue;
 		}
 
@@ -1078,7 +1124,36 @@ roff_text(ROFFCALL_ARGS)
 		break;
 	}
 
-	return((*tree->cb.roffout)(tree->arg, tok));
+	if ( ! (*tree->cb.roffout)(tree->arg, tok))
+		return(0);
+
+	/* 
+	 * If we're the first parser (*argv == tree->cur) then purge out
+	 * any additional punctuation, should there be any remaining at
+	 * the end of line. 
+	 */
+
+	if (first && *argv) {
+		i = 0;
+		while (argv[i])
+			i++;
+
+		assert(i > 0);
+		if ( ! roffispunct(argv[--i]))
+			return(1);
+
+		while (i >= 0 && roffispunct(argv[i]))
+			i--;
+
+		assert(0 != i);
+		i++;
+
+		while (argv[i])
+			if ( ! (*tree->cb.roffdata)(tree->arg, 0, argv[i++]))
+				return(0);
+	}
+
+	return(1);
 }
 
 
diff --git a/validate.c b/validate.c
index e5ca3a50..e8f9cec8 100644
--- a/validate.c
+++ b/validate.c
@@ -52,7 +52,7 @@ static	void		 roffmsg(void *arg, enum roffmsg,
 static	int		 roffhead(void *);
 static	int		 rofftail(void *);
 static	int		 roffin(void *, int, int *, char **);
-static	int		 roffdata(void *, char *);
+static	int		 roffdata(void *, int, char *);
 static	int		 roffout(void *, int);
 static	int		 roffblkin(void *, int, int *, char **);
 static	int		 roffblkout(void *, int);
@@ -60,7 +60,7 @@ static	int		 roffspecial(void *, int);
 
 static	int		 mbuf_newline(struct md_valid *);
 static	int		 mbuf_indent(struct md_valid *);
-static	int		 mbuf_data(struct md_valid *, char *);
+static	int		 mbuf_data(struct md_valid *, int, char *);
 
 
 static int
@@ -92,14 +92,11 @@ mbuf_newline(struct md_valid *p)
 
 
 static int
-mbuf_data(struct md_valid *p, char *buf)
+mbuf_data(struct md_valid *p, int space, char *buf)
 {
-	int		 space;
 	size_t		 sz;
 	char		*bufp;
 
-	space = 1; /* FIXME */
-
 	assert(p->mbuf);
 	assert(0 != p->indent);
 
@@ -227,13 +224,15 @@ roffhead(void *arg)
 	assert(arg);
 	p = (struct md_valid *)arg;
 
-	if ( ! md_buf_putstring(p->mbuf, "BEGIN"))
+	if ( ! md_buf_putstring(p->mbuf, "<?xml version=\"1.0\" "
+				"encoding=\"UTF-8\"?>\n"))
 		return(0);
-	p->indent++;
-	if ( ! mbuf_newline(p))
+
+	if ( ! md_buf_putstring(p->mbuf, "<mdoc>"))
 		return(0);
+	p->indent++;
 
-	return(1);
+	return(mbuf_newline(p));
 }
 
 
@@ -247,10 +246,7 @@ rofftail(void *arg)
 
 	if (0 != p->pos && ! mbuf_newline(p))
 		return(0);
-
-	if ( ! md_buf_putstring(p->mbuf, "END\n"))
-		return(0);
-	return(1);
+	return(md_buf_putstring(p->mbuf, "</mdoc>\n"));
 }
 
 
@@ -334,6 +330,7 @@ roffin(void *arg, int tok, int *argcp, char **argvp)
 	if (0 == p->pos && ! mbuf_indent(p))
 		return(0);
 
+	/* FIXME: not always with a space... */
 	if ( ! md_buf_putstring(p->mbuf, " <"))
 		return(0);
 	if ( ! md_buf_putstring(p->mbuf, toknames[tok]))
@@ -406,11 +403,11 @@ roffmsg(void *arg, enum roffmsg lvl,
 
 
 static int
-roffdata(void *arg, char *buf)
+roffdata(void *arg, int space, char *buf)
 {
 	struct md_valid	*p;
 
 	assert(arg);
 	p = (struct md_valid *)arg;
-	return(mbuf_data(p, buf));
+	return(mbuf_data(p, space, buf));
 }