Improvements to the xml part.

6 files changed, 213 insertions, 113 deletions

diff --git a/Makefile b/Makefile
index fb85076f..e659dbc0 100644
--- a/Makefile
+++ b/Makefile
@@ -25,7 +25,7 @@ FAIL	= test.0 test.1 test.2 test.3 test.4 test.5 test.6 \
 
 SUCCEED	= test.7 test.8 test.9 test.10 test.11 test.12 test.13 \
 	  test.14 test.16 test.17 test.18 test.19 test.21 test.23 \
-	  test.25 test.28 test.29 test.31
+	  test.25 test.28 test.29 test.31 test.32 test.33 test.34
 
 
 all: mdocml
diff --git a/libmdocml.c b/libmdocml.c
index a10519bd..7cf6c079 100644
--- a/libmdocml.c
+++ b/libmdocml.c
@@ -150,7 +150,7 @@ md_run_leave(const struct md_args *args, struct md_mbuf *mbuf,
 	}
 
 	/* Make final flush of buffer. */
-	if ( ! md_buf_flush(mbuf))
+	if (-1 != c && ! md_buf_flush(mbuf))
 		return(-1);
 
 	return(c);
diff --git a/mdocml.1 b/mdocml.1
index 765b3a41..4bfd0cf3 100644
--- a/mdocml.1
+++ b/mdocml.1
@@ -84,7 +84,6 @@ structured ones:
 The engine doesn't understand the
 .Sq \&Xo ,
 .Sq \&Xc ,
-.Sq \&Ns ,
 .Sq \&No ,
 .Sq \&Db ,
 .Sq \&Sm ,
@@ -102,4 +101,9 @@ If terminating punctuation is found, then
 remaining tokens are flushed after line scope is closed, not just the
 last one.
 .El
+.Pp
+The roff engine in 
+.Nm
+produces text in-line; thus, output may already be partially written by
+the time an error is encountered.
 .\" .Sh BUGS
diff --git a/mdocml.c b/mdocml.c
index 4f7eac9a..84ceec26 100644
--- a/mdocml.c
+++ b/mdocml.c
@@ -106,6 +106,9 @@ leave_io(const struct md_buf *out,
 		warn("%s", out->name);
 		c = 1;
 	}
+	if (1 == c && STDOUT_FILENO != out->fd)
+		if (-1 == unlink(out->name))
+			warn("%s", out->name);
 
 	return(c);
 }
diff --git a/roff.c b/roff.c
index cb5983b5..9b513050 100644
--- a/roff.c
+++ b/roff.c
@@ -112,6 +112,7 @@ struct	rofftree {
 static	int		  roff_Dd(ROFFCALL_ARGS);
 static	int		  roff_Dt(ROFFCALL_ARGS);
 static	int		  roff_Os(ROFFCALL_ARGS);
+static	int		  roff_Ns(ROFFCALL_ARGS);
 
 static	int		  roff_layout(ROFFCALL_ARGS);
 static	int		  roff_text(ROFFCALL_ARGS);
@@ -126,6 +127,7 @@ static	void		  roff_warn(const struct rofftree *,
 static	void		  roff_err(const struct rofftree *, 
 				const char *, char *, ...);
 
+static	int		  roffpurgepunct(struct rofftree *, char **);
 static	int		  roffscan(int, const int *);
 static	int		  rofffindtok(const char *);
 static	int		  rofffindarg(const char *);
@@ -253,7 +255,7 @@ static	const struct rofftok tokens[ROFF_MAX] = {
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Fx */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ms */
 	{   NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* No */
-	{        NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ns */
+	{     roff_Ns, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ns */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Nx */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ox */
 	{   roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pc */
@@ -878,6 +880,29 @@ roffnextopt(const struct rofftree *tree, int tok,
 
 
 static int
+roffpurgepunct(struct rofftree *tree, char **argv)
+{
+	int		 i;
+
+	i = 0;
+	while (argv[i])
+		i++;
+	assert(i > 0);
+	if ( ! roffispunct(argv[--i]))
+		return(1);
+	while (i >= 0 && roffispunct(argv[i]))
+		i--;
+	i++;
+
+	/* LINTED */
+	while (argv[i])
+		if ( ! (*tree->cb.roffdata)(tree->arg, 0, argv[i++]))
+			return(0);
+	return(1);
+}
+
+
+static int
 roffparseopts(struct rofftree *tree, int tok, 
 		char ***args, int *argc, char **argv)
 {
@@ -1047,6 +1072,79 @@ roff_Dt(ROFFCALL_ARGS)
 
 /* ARGSUSED */
 static	int
+roff_Ns(ROFFCALL_ARGS)
+{
+	int		 j, c, first;
+
+	first = (*argv == tree->cur);
+
+	argv++;
+
+	if (ROFF_MAX != (c = rofffindcallable(*argv))) {
+		if (NULL == tokens[c].cb) {
+			roff_err(tree, *argv, "unsupported macro `%s'",
+					toknames[c]);
+			return(0);
+		}
+		if ( ! (*tree->cb.roffspecial)(tree->arg, tok))
+			return(0);
+		if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER))
+			return(0);
+		if ( ! first)
+			return(1);
+		return(roffpurgepunct(tree, argv));
+	} 
+
+	if ( ! (*tree->cb.roffdata)(tree->arg, 0, *argv++))
+		return(0);
+
+	while (*argv) {
+		if (ROFF_MAX == (c = rofffindcallable(*argv))) {
+			if ( ! roffispunct(*argv)) {
+				if ( ! (*tree->cb.roffdata)
+						(tree->arg, 1, *argv++))
+					return(0);
+				continue;
+			}
+			
+			/* FIXME: this is identical to that of
+			 * roff_text. */
+
+			/* See if only punctuation remains. */
+
+			for (j = 0; argv[j]; j++)
+				if ( ! roffispunct(argv[j]))
+					break;
+
+			if (argv[j]) {
+				if ( ! (*tree->cb.roffdata)
+						(tree->arg, 0, *argv++))
+					return(0);
+				continue;
+			}
+
+			/*  Only punctuation remains. */
+
+			break;
+		}
+		if (NULL == tokens[c].cb) {
+			roff_err(tree, *argv, "unsupported macro `%s'",
+					toknames[c]);
+			return(0);
+		}
+		if ( ! (*tokens[c].cb)(c, tree, argv, ROFF_ENTER))
+			return(0);
+		break;
+	}
+
+	if ( ! first)
+		return(1);
+	return(roffpurgepunct(tree, argv));
+}
+
+
+/* ARGSUSED */
+static	int
 roff_Os(ROFFCALL_ARGS)
 {
 	char		*p;
@@ -1188,23 +1286,8 @@ roff_layout(ROFFCALL_ARGS)
 	 * a token isn't punctuation.
 	 */
 
-	i = 0;
-	while (argv[i])
-		i++;
-
-	assert(i > 0);
-	if ( ! roffispunct(argv[--i]))
-		return((*tree->cb.roffout)(tree->arg, tok));
-
-	while (i >= 0 && roffispunct(argv[i]))
-		i--;
-
-	i++;
-
-	/* LINTED */
-	while (argv[i])
-		if ( ! (*tree->cb.roffdata)(tree->arg, 0, argv[i++]))
-			return(0);
+	if ( ! roffpurgepunct(tree, argv))
+		return(0);
 
 	return((*tree->cb.roffout)(tree->arg, tok));
 }
@@ -1314,29 +1397,7 @@ roff_text(ROFFCALL_ARGS)
 	if ( ! first)
 		return(1);
 
-	/*
-	 * We're the line-dominant macro.  Check if there's remaining
-	 * punctuation.  If there is, then flush it out before exiting.
-	 */
-
-	i = 0;
-	while (argv[i])
-		i++;
-
-	assert(i > 0);
-	if ( ! roffispunct(argv[--i]))
-		return(1);
-
-	while (i >= 0 && roffispunct(argv[i]))
-		i--;
-	i++;
-
-	/* LINTED */
-	while (argv[i])
-		if ( ! (*tree->cb.roffdata)(tree->arg, 0, argv[i++]))
-			return(0);
-
-	return(1);
+	return(roffpurgepunct(tree, argv));
 }
 
 
diff --git a/xml.c b/xml.c
index 99f41f07..4445f9c2 100644
--- a/xml.c
+++ b/xml.c
@@ -29,12 +29,18 @@
 #include "private.h"
 
 #define	MAXINDENT	 8
-#define	COLUMNS		 60
+#define	COLUMNS		 72
 
 #ifdef	__linux__ /* FIXME */
 #define	strlcat		 strncat
 #endif
 
+enum	md_ns {
+	MD_NS_BLOCK,
+	MD_NS_INLINE,
+	MD_NS_DEFAULT
+};
+
 enum	md_tok {
 	MD_BLKIN,
 	MD_BLKOUT,
@@ -78,6 +84,76 @@ static	int		 mbuf_nputstring(struct md_xml *,
 static	int		 mbuf_puts(struct md_xml *, const char *);
 static	int		 mbuf_nputs(struct md_xml *, 
 				const char *, size_t);
+static	int		 mbuf_begintag(struct md_xml *, const char *, 
+				enum md_ns, int *, char **);
+static	int		 mbuf_endtag(struct md_xml *, 
+				const char *, enum md_ns);
+
+
+static int
+mbuf_begintag(struct md_xml *p, const char *name, enum md_ns ns, 
+		int *argc, char **argv)
+{
+	int		 i;
+
+	if ( ! mbuf_nputs(p, "<", 1))
+		return(0);
+
+	switch (ns) {
+		case (MD_NS_BLOCK):
+			if ( ! mbuf_nputs(p, "block:", 6))
+				return(0);
+			break;
+		case (MD_NS_INLINE):
+			if ( ! mbuf_nputs(p, "inline:", 7))
+				return(0);
+			break;
+		default:
+			break;
+	}
+
+	if ( ! mbuf_puts(p, name))
+		return(0);
+
+	for (i = 0; ROFF_ARGMAX != argc[i]; i++) {
+		if ( ! mbuf_nputs(p, " ", 1))
+			return(0);
+		if ( ! mbuf_puts(p, tokargnames[argc[i]]))
+			return(0);
+		if ( ! mbuf_nputs(p, "=\"", 2))
+			return(0);
+		if ( ! mbuf_putstring(p, argv[i] ? argv[i] : "true"))
+			return(0);
+		if ( ! mbuf_nputs(p, "\"", 1))
+			return(0);
+	}
+	return(mbuf_nputs(p, ">", 1));
+}
+
+
+static int
+mbuf_endtag(struct md_xml *p, const char *tag, enum md_ns ns)
+{
+	if ( ! mbuf_nputs(p, "</", 2))
+		return(0);
+
+	switch (ns) {
+		case (MD_NS_BLOCK):
+			if ( ! mbuf_nputs(p, "block:", 6))
+				return(0);
+			break;
+		case (MD_NS_INLINE):
+			if ( ! mbuf_nputs(p, "inline:", 7))
+				return(0);
+			break;
+		default:
+			break;
+	}
+
+	if ( ! mbuf_puts(p, tag))
+		return(0);
+	return(mbuf_nputs(p, ">", 1));
+}
 
 
 static int
@@ -105,6 +181,16 @@ mbuf_nputstring(struct md_xml *p, const char *buf, size_t sz)
 				return(0);
 			p->pos += 6;
 			break;
+		case ('<'):
+			if ( ! md_buf_puts(p->mbuf, "&lt;", 4))
+				return(0);
+			p->pos += 4;
+			break;
+		case ('>'):
+			if ( ! md_buf_puts(p->mbuf, "&gt;", 4))
+				return(0);
+			p->pos += 4;
+			break;
 		default:
 			if ( ! md_buf_putchar(p->mbuf, buf[i]))
 				return(0);
@@ -318,15 +404,14 @@ rofftail(void *arg)
 	if (0 != p->pos && ! mbuf_newline(p))
 		return(0);
 
-	if ( ! mbuf_puts(p, "</mdoc>"))
+	p->last = MD_BLKOUT;
+	if ( ! mbuf_endtag(p, "mdoc", MD_NS_DEFAULT))
 		return(0);
 
-	p->last = MD_BLKOUT;
 	return(mbuf_newline(p));
 }
 
 
-/* ARGSUSED */
 static int
 roffspecial(void *arg, int tok)
 {
@@ -335,6 +420,8 @@ roffspecial(void *arg, int tok)
 	assert(arg);
 	p = (struct md_xml *)arg;
 
+	/* FIXME: this is completely ad hoc. */
+
 	switch (tok) {
 	case (ROFF_Ns):
 		p->last = MD_OVERRIDE;
@@ -351,7 +438,6 @@ static int
 roffblkin(void *arg, int tok, int *argc, char **argv)
 {
 	struct md_xml	*p;
-	int		 i;
 
 	assert(arg);
 	p = (struct md_xml *)arg;
@@ -364,33 +450,14 @@ roffblkin(void *arg, int tok, int *argc, char **argv)
 	} else if ( ! mbuf_indent(p))
 		return(0);
 
-	if ( ! mbuf_nputs(p, "<", 1))
-		return(0);
-	if ( ! mbuf_nputs(p, "block:", 6))
-		return(0);
-	if ( ! mbuf_puts(p, toknames[tok]))
-		return(0);
-
 	/* FIXME: xml won't like standards args (e.g., p1003.1-90). */
 
-	for (i = 0; ROFF_ARGMAX != argc[i]; i++) {
-		if ( ! mbuf_nputs(p, " ", 1))
-			return(0);
-		if ( ! mbuf_puts(p, tokargnames[argc[i]]))
-			return(0);
-		if ( ! mbuf_nputs(p, "=\"", 2))
-			return(0);
-		if ( ! mbuf_putstring(p, argv[i] ? argv[i] : "true"))
-			return(0);
-		if ( ! mbuf_nputs(p, "\"", 1))
-			return(0);
-	}
-
-	if ( ! mbuf_nputs(p, ">", 1))
-		return(0);
-
 	p->last = MD_BLKIN;
 	p->indent++;
+
+	if ( ! mbuf_begintag(p, toknames[tok], MD_NS_BLOCK,
+				argc, argv))
+		return(0);
 	return(mbuf_newline(p));
 }
 
@@ -413,16 +480,10 @@ roffblkout(void *arg, int tok)
 	} else if ( ! mbuf_indent(p))
 		return(0);
 
-	if ( ! mbuf_nputs(p, "</", 2))
-		return(0);
-	if ( ! mbuf_nputs(p, "block:", 6))
-		return(0);
-	if ( ! mbuf_puts(p, toknames[tok]))
-		return(0);
-	if ( ! mbuf_nputs(p, ">", 1))
-		return(0);
-
 	p->last = MD_BLKOUT;
+
+	if ( ! mbuf_endtag(p, toknames[tok], MD_NS_BLOCK))
+		return(0);
 	return(mbuf_newline(p));
 }
 
@@ -431,7 +492,6 @@ static int
 roffin(void *arg, int tok, int *argc, char **argv)
 {
 	struct md_xml	*p;
-	int		 i;
 
 	assert(arg);
 	p = (struct md_xml *)arg;
@@ -461,27 +521,8 @@ roffin(void *arg, int tok, int *argc, char **argv)
 		return(0);
 
 	p->last = MD_IN;
-
-	if ( ! mbuf_nputs(p, "<", 1))
-		return(0);
-	if ( ! mbuf_nputs(p, "inline:", 7))
-		return(0);
-	if ( ! mbuf_puts(p, toknames[tok]))
-		return(0);
-
-	for (i = 0; ROFF_ARGMAX != argc[i]; i++) {
-		if ( ! mbuf_nputs(p, " ", 1))
-			return(0);
-		if ( ! mbuf_puts(p, tokargnames[argc[i]]))
-			return(0);
-		if ( ! mbuf_nputs(p, "=\"", 2))
-			return(0);
-		if ( ! mbuf_putstring(p, argv[i] ? argv[i] : "true"))
-			return(0);
-		if ( ! mbuf_nputs(p, "\"", 1))
-			return(0);
-	}
-	return(mbuf_nputs(p, ">", 1));
+	return(mbuf_begintag(p, toknames[tok], 
+				MD_NS_INLINE, argc, argv));
 }
 
 
@@ -493,20 +534,11 @@ roffout(void *arg, int tok)
 	assert(arg);
 	p = (struct md_xml *)arg;
 
-	/* Continue with a regular out token. */
-
 	if (0 == p->pos && ! mbuf_indent(p))
 		return(0);
 
 	p->last = MD_OUT;
-
-	if ( ! mbuf_nputs(p, "</", 2))
-		return(0);
-	if ( ! mbuf_nputs(p, "inline:", 7))
-		return(0);
-	if ( ! mbuf_puts(p, toknames[tok]))
-		return(0);
-	return(mbuf_nputs(p, ">", 1));
+	return(mbuf_endtag(p, toknames[tok], MD_NS_INLINE));
 }