Preliminary xml output filter (validate-renamed) done.

7 files changed, 147 insertions, 99 deletions

diff --git a/Makefile b/Makefile
index 0dfdb46a..b138633b 100644
--- a/Makefile
+++ b/Makefile
@@ -2,15 +2,15 @@ CFLAGS += -W -Wall -Wno-unused-parameter -g
 
 LINTFLAGS += -c -e -f -u
 
-LNS	= mdocml.ln html4_strict.ln validate.ln libmdocml.ln roff.ln
+LNS	= mdocml.ln html4_strict.ln xml.ln libmdocml.ln roff.ln
 
 LLNS	= llib-lmdocml.ln
 
 LIBS	= libmdocml.a
 
-OBJS	= mdocml.o html4_strict.o validate.o libmdocml.o roff.o
+OBJS	= mdocml.o html4_strict.o xml.o libmdocml.o roff.o
 
-SRCS	= mdocml.c html4_strict.c validate.c libmdocml.c roff.c
+SRCS	= mdocml.c html4_strict.c xml.c libmdocml.c roff.c
 
 HEADS	= libmdocml.h private.h
 
@@ -49,19 +49,19 @@ mdocml.tgz: $(INSTALL)
 	( cd .dist/ && tar zcf ../mdocml.tgz mdocml/ )
 	rm -rf .dist/
 
-llib-lmdocml.ln: mdocml.ln libmdocml.ln html4_strict.ln validate.ln roff.ln
-	$(LINT) $(LINTFLAGS) -Cmdocml mdocml.ln libmdocml.ln html4_strict.ln validate.ln roff.ln
+llib-lmdocml.ln: mdocml.ln libmdocml.ln html4_strict.ln xml.ln roff.ln
+	$(LINT) $(LINTFLAGS) -Cmdocml mdocml.ln libmdocml.ln html4_strict.ln xml.ln roff.ln
 
 mdocml.ln: mdocml.c libmdocml.h
 
 mdocml.o: mdocml.c libmdocml.h
 
-libmdocml.a: libmdocml.o html4_strict.o validate.o roff.o
-	$(AR) rs $@ libmdocml.o html4_strict.o validate.o roff.o
+libmdocml.a: libmdocml.o html4_strict.o xml.o roff.o
+	$(AR) rs $@ libmdocml.o html4_strict.o xml.o roff.o
 
-validate.ln: validate.c private.h libmdocml.h
+xml.ln: xml.c private.h libmdocml.h
 
-validate.o: validate.c private.h libmdocml.h
+xml.o: xml.c private.h libmdocml.h
 
 html4_strict.ln: html4_strict.c private.h libmdocml.h
 
diff --git a/libmdocml.c b/libmdocml.c
index 11eff674..a10519bd 100644
--- a/libmdocml.c
+++ b/libmdocml.c
@@ -144,7 +144,7 @@ md_run_leave(const struct md_args *args, struct md_mbuf *mbuf,
 			c = -1;
 		break;
 	default:
-		if ( ! md_exit_valid(data, -1 == c ? 0 : 1))
+		if ( ! md_exit_xml(data, -1 == c ? 0 : 1))
 			c = -1;
 		break;
 	}
@@ -176,7 +176,7 @@ md_run_enter(const struct md_args *args, struct md_mbuf *mbuf,
 		fp = md_line_html4_strict;
 		break;
 	default:
-		fp = md_line_valid;
+		fp = md_line_xml;
 		break;
 	}
 
@@ -247,8 +247,7 @@ md_run(const struct md_args *args,
 			(args, &mbuf, &rbuf);
 		break;
 	default:
-		data = md_init_valid
-			(args, &mbuf, &rbuf);
+		data = md_init_xml(args, &mbuf, &rbuf);
 		break;
 	}
 
diff --git a/libmdocml.h b/libmdocml.h
index c2576146..74174e5f 100644
--- a/libmdocml.h
+++ b/libmdocml.h
@@ -21,7 +21,7 @@
 
 #include <sys/types.h>
 
-struct 	md_params_valid {
+struct 	md_params_xml {
 	int		 dummy;
 };
 
@@ -30,12 +30,12 @@ struct 	md_params_html4_strict {
 };
 
 union	md_params {
-	struct md_params_valid valid;
+	struct md_params_xml xml;
 	struct md_params_html4_strict html4_strict;
 };
 
 enum	md_type {
-	MD_VALID,		/* Validate input only. */
+	MD_XML,			/* XML. */
 	MD_HTML4_STRICT		/* HTML4.01-strict. */
 };
 
diff --git a/mdocml.1 b/mdocml.1
index c5f86d66..83a036d7 100644
--- a/mdocml.1
+++ b/mdocml.1
@@ -13,7 +13,7 @@
 .\"
 .Sh SYNOPSIS
 .Nm mdocml
-.Op Fl vW
+.Op Fl W
 .Op Fl f Ar filter
 .Op Fl o Ar outfile
 .Op Ar infile
@@ -27,32 +27,49 @@ formatted manual source and passes results into the output filter
 dictated by 
 .Fl f Ar filter .
 The only current output filter is
-.Ar validate ,
+.Ar xml ,
 the default.  The arguments are as follows:
-.Bl -tag -width "\-o outfile,"
-.It Sq Fl o Ar outfile ,
+.Bl -tag -width "\-o outfile"
+.It Fl f Ar filter
+The output filter name, which defaults to
+.Ar xml .
+.It Fl o Ar outfile
 Place output in 
 .Ar outfile ,
 which may be
 .Qq \-
-for standard output.  The default is standard output.
+for stdout.  The default is stdout.
+.It Fl W
+Print compiler warnings to stderr.
 .It Ar infile
 Read input from
 .Ar infile ,
 which may be 
 .Qq \-
-for standard input.  The default is standard input.
-.It Fl v 
-Print messages (output depends on filter).
-.It Fl W
-Print warnings (output depends on filter).
+for stdin.  The default is stdin.
 .El
-.Ss Validate Filter
-The validate filter, specified by
-.Fl f Ar validate ,
-is the default filter.  It produces no output, so specifying
-.Fl o Ar outfile
-results in an empty file.
+.Ss XML Filter
+The XML filter, specified by
+.Fl f Ar xml ,
+is the default filter.  It creates an XML document where element names are
+their respective roff macro names.  Each element name has an associated
+namespace, which is one of 
+.Qq block ,
+.Qq inline ,
+or
+.Qq special ,
+corresponding to the display mode of a node.
+.Pp
+Sample output follows:
+.Bd -literal
+<?xml version="1.0" encoding="UTF-8"?>
+<block:mdoc>
+  <block:Sh>
+    <inline:Sh>NAME</inline:Sh> <inline:Nm>example</inline:Nm>
+    <inline:Nd>example text</inline:Nd>
+  </block:Sh>
+</block:mdoc>
+.Ed
 .\" 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
diff --git a/private.h b/private.h
index 95d8a41d..7811e843 100644
--- a/private.h
+++ b/private.h
@@ -213,7 +213,7 @@ struct	roffcb {
 			const char *, const char *, char *);
 	int	(*roffhead)(void *);
 	int	(*rofftail)(void *);
-	int	(*roffin)(void *, int, int, int *, char **);
+	int	(*roffin)(void *, int, int *, char **);
 	int	(*roffdata)(void *, int, char *);
 	int	(*roffout)(void *, int);
 	int	(*roffblkin)(void *, int, int *, char **);
@@ -233,10 +233,10 @@ void		 *md_init_html4_strict(const struct md_args *,
 int		  md_line_html4_strict(void *, char *);
 int		  md_exit_html4_strict(void *, int);
 
-void		 *md_init_valid(const struct md_args *,
+void		 *md_init_xml(const struct md_args *,
 			struct md_mbuf *, const struct md_rbuf *);
-int		  md_line_valid(void *, char *);
-int		  md_exit_valid(void *, int);
+int		  md_line_xml(void *, char *);
+int		  md_exit_xml(void *, int);
 
 int	 	  md_buf_puts(struct md_mbuf *, const char *, size_t);
 int	 	  md_buf_putchar(struct md_mbuf *, char);
diff --git a/roff.c b/roff.c
index a1f5f660..7d88c8a1 100644
--- a/roff.c
+++ b/roff.c
@@ -979,7 +979,7 @@ roff_layout(ROFFCALL_ARGS)
 	if (NULL == *argv)
 		return(1);
 
-	if ( ! (*tree->cb.roffin)(tree->arg, tok, 0, argcp, argvp))
+	if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp))
 		return(0);
 
 	if ( ! (ROFF_PARSED & tokens[tok].flags)) {
@@ -1075,7 +1075,7 @@ roff_text(ROFFCALL_ARGS)
 	argcp[i] = ROFF_ARGMAX;
 	argvp[i] = NULL;
 
-	if ( ! (*tree->cb.roffin)(tree->arg, tok, 1, argcp, argvp))
+	if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp))
 		return(0);
 
 	if ( ! (ROFF_PARSED & tokens[tok].flags)) {
diff --git a/validate.c b/xml.c
index efd3bddf..d9916767 100644
--- a/validate.c
+++ b/xml.c
@@ -29,7 +29,7 @@
 #include "private.h"
 
 #define	INDENT		 4
-#define	COLUMNS		 72
+#define	COLUMNS		 60
 
 #ifdef	__linux__ /* FIXME */
 #define	strlcat		 strncat
@@ -43,7 +43,7 @@ enum	md_tok {
 	MD_TEXT
 };
 
-struct	md_valid {
+struct	md_xml {
 	const struct md_args	*args;
 	const struct md_rbuf	*rbuf;
 
@@ -60,24 +60,24 @@ static	void		 roffmsg(void *arg, enum roffmsg,
 				const char *, const char *, char *);
 static	int		 roffhead(void *);
 static	int		 rofftail(void *);
-static	int		 roffin(void *, int, int, int *, char **);
+static	int		 roffin(void *, int, int *, char **);
 static	int		 roffdata(void *, int, char *);
 static	int		 roffout(void *, int);
 static	int		 roffblkin(void *, int, int *, char **);
 static	int		 roffblkout(void *, int);
 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 *, int, char *);
-static	int		 mbuf_putstring(struct md_valid *, 
+static	int		 mbuf_newline(struct md_xml *);
+static	int		 mbuf_indent(struct md_xml *);
+static	int		 mbuf_data(struct md_xml *, int, char *);
+static	int		 mbuf_putstring(struct md_xml *, 
 				const char *);
-static	int		 mbuf_nputstring(struct md_valid *, 
+static	int		 mbuf_nputstring(struct md_xml *, 
 				const char *, size_t);
 
 
 static int
-mbuf_putstring(struct md_valid *p, const char *buf)
+mbuf_putstring(struct md_xml *p, const char *buf)
 {
 
 	return(mbuf_nputstring(p, buf, strlen(buf)));
@@ -85,7 +85,7 @@ mbuf_putstring(struct md_valid *p, const char *buf)
 
 
 static int
-mbuf_nputstring(struct md_valid *p, const char *buf, size_t sz)
+mbuf_nputstring(struct md_xml *p, const char *buf, size_t sz)
 {
 
 	p->pos += sz;
@@ -94,7 +94,7 @@ mbuf_nputstring(struct md_valid *p, const char *buf, size_t sz)
 
 
 static int
-mbuf_indent(struct md_valid *p)
+mbuf_indent(struct md_xml *p)
 {
 	size_t		 i;
 
@@ -111,7 +111,7 @@ mbuf_indent(struct md_valid *p)
 
 
 static int
-mbuf_newline(struct md_valid *p)
+mbuf_newline(struct md_xml *p)
 {
 
 	if ( ! md_buf_putchar(p->mbuf, '\n'))
@@ -123,7 +123,7 @@ mbuf_newline(struct md_valid *p)
 
 
 static int
-mbuf_data(struct md_valid *p, int space, char *buf)
+mbuf_data(struct md_xml *p, int space, char *buf)
 {
 	size_t		 sz;
 	char		*bufp;
@@ -131,11 +131,6 @@ mbuf_data(struct md_valid *p, int space, char *buf)
 	assert(p->mbuf);
 	assert(0 != p->indent);
 
-	/* 
-	 * FIXME: punctuation/no-space stuff shouldn't have a newline
-	 * before it.
-	 */
-
 	if (MD_LITERAL & p->flags)
 		return(mbuf_putstring(p, buf));
 
@@ -168,17 +163,22 @@ mbuf_data(struct md_valid *p, int space, char *buf)
 			continue;
 		}
 
-		if (sz + p->pos >= COLUMNS) {
+		if (space && sz + p->pos >= COLUMNS) {
 			if ( ! mbuf_newline(p))
 				return(0);
 			if ( ! mbuf_indent(p))
 				return(0);
-		} else if (space) 
+		} else if (space) {
 			if ( ! mbuf_nputstring(p, " ", 1))
 				return(0);
+		}
 
 		if ( ! mbuf_nputstring(p, bufp, sz))
 			return(0);
+
+		if ( ! space && p->pos >= COLUMNS)
+			if ( ! mbuf_newline(p))
+				return(0);
 	}
 
 	return(1);
@@ -186,22 +186,22 @@ mbuf_data(struct md_valid *p, int space, char *buf)
 
 
 int
-md_line_valid(void *arg, char *buf)
+md_line_xml(void *arg, char *buf)
 {
-	struct md_valid	*p;
+	struct md_xml	*p;
 
-	p = (struct md_valid *)arg;
+	p = (struct md_xml *)arg;
 	return(roff_engine(p->tree, buf));
 }
 
 
 int
-md_exit_valid(void *data, int flush)
+md_exit_xml(void *data, int flush)
 {
 	int		 c;
-	struct md_valid	*p;
+	struct md_xml	*p;
 
-	p = (struct md_valid *)data;
+	p = (struct md_xml *)data;
 	c = roff_free(p->tree, flush);
 	free(p);
 
@@ -210,11 +210,11 @@ md_exit_valid(void *data, int flush)
 
 
 void *
-md_init_valid(const struct md_args *args,
+md_init_xml(const struct md_args *args,
 		struct md_mbuf *mbuf, const struct md_rbuf *rbuf)
 {
 	struct roffcb	 cb;
-	struct md_valid	*p;
+	struct md_xml	*p;
 
 	cb.roffhead = roffhead;
 	cb.rofftail = rofftail;
@@ -226,7 +226,7 @@ md_init_valid(const struct md_args *args,
 	cb.roffmsg = roffmsg;
 	cb.roffdata = roffdata;
 
-	if (NULL == (p = calloc(1, sizeof(struct md_valid))))
+	if (NULL == (p = calloc(1, sizeof(struct md_xml))))
 		err(1, "malloc");
 
 	p->args = args;
@@ -248,18 +248,19 @@ md_init_valid(const struct md_args *args,
 static int
 roffhead(void *arg)
 {
-	struct md_valid	*p;
+	struct md_xml	*p;
 
 	assert(arg);
-	p = (struct md_valid *)arg;
+	p = (struct md_xml *)arg;
 
 	if ( ! mbuf_putstring(p, "<?xml version=\"1.0\" "
 				"encoding=\"UTF-8\"?>\n"))
 		return(0);
-	if ( ! mbuf_nputstring(p, "<mdoc>", 6))
+	if ( ! mbuf_nputstring(p, "<block:mdoc>", 12))
 		return(0);
 
 	p->indent++;
+	p->last = MD_BLKIN;
 	return(mbuf_newline(p));
 }
 
@@ -267,16 +268,18 @@ roffhead(void *arg)
 static int
 rofftail(void *arg)
 {
-	struct md_valid	*p;
+	struct md_xml	*p;
 
 	assert(arg);
-	p = (struct md_valid *)arg;
+	p = (struct md_xml *)arg;
 
 	if (0 != p->pos && ! mbuf_newline(p))
 		return(0);
 
-	if ( ! mbuf_nputstring(p, "</mdoc>", 7))
+	if ( ! mbuf_nputstring(p, "</block:mdoc>", 13))
 		return(0);
+
+	p->last = MD_BLKOUT;
 	return(mbuf_newline(p));
 }
 
@@ -286,6 +289,7 @@ static int
 roffspecial(void *arg, int tok)
 {
 
+	/* FIXME */
 	return(1);
 }
 
@@ -293,11 +297,11 @@ roffspecial(void *arg, int tok)
 static int
 roffblkin(void *arg, int tok, int *argc, char **argv)
 {
-	struct md_valid	*p;
+	struct md_xml	*p;
 	int		 i;
 
 	assert(arg);
-	p = (struct md_valid *)arg;
+	p = (struct md_xml *)arg;
 
 	if (0 != p->pos) {
 		if ( ! mbuf_newline(p))
@@ -309,6 +313,8 @@ roffblkin(void *arg, int tok, int *argc, char **argv)
 
 	if ( ! mbuf_nputstring(p, "<", 1))
 		return(0);
+	if ( ! mbuf_nputstring(p, "block:", 6))
+		return(0);
 	if ( ! mbuf_putstring(p, toknames[tok]))
 		return(0);
 
@@ -327,21 +333,20 @@ roffblkin(void *arg, int tok, int *argc, char **argv)
 
 	if ( ! mbuf_nputstring(p, ">", 1))
 		return(0);
-	if ( ! mbuf_newline(p))
-		return(0);
 
+	p->last = MD_BLKIN;
 	p->indent++;
-	return(1);
+	return(mbuf_newline(p));
 }
 
 
 static int
 roffblkout(void *arg, int tok)
 {
-	struct md_valid	*p;
+	struct md_xml	*p;
 
 	assert(arg);
-	p = (struct md_valid *)arg;
+	p = (struct md_xml *)arg;
 
 	p->indent--;
 
@@ -355,37 +360,57 @@ roffblkout(void *arg, int tok)
 
 	if ( ! mbuf_nputstring(p, "</", 2))
 		return(0);
+	if ( ! mbuf_nputstring(p, "block:", 6))
+		return(0);
 	if ( ! mbuf_putstring(p, toknames[tok]))
 		return(0);
 	if ( ! mbuf_nputstring(p, ">", 1))
 		return(0);
-	if ( ! mbuf_newline(p))
-		return(0);
 
-	return(1);
+	p->last = MD_BLKOUT;
+	return(mbuf_newline(p));
 }
 
 
 static int
-roffin(void *arg, int tok, int space, int *argc, char **argv)
+roffin(void *arg, int tok, int *argc, char **argv)
 {
-	struct md_valid	*p;
+	struct md_xml	*p;
 	int		 i;
 
 	assert(arg);
-	p = (struct md_valid *)arg;
-
-	if (0 == p->pos && ! mbuf_indent(p))
-		return(0);
+	p = (struct md_xml *)arg;
 
 	/* 
-	 * FIXME: put into a buffer before writing (check line length).
+	 * FIXME: put all of this in a buffer, then check the buffer
+	 * length versus the column width for nicer output.  This is a
+	 * bit hacky.
 	 */
 
-	if (space && ! mbuf_nputstring(p, " ", 1))
+	if (p->pos + 11 > COLUMNS) 
+		if ( ! mbuf_newline(p))
+			return(0);
+
+	if (0 != p->pos) {
+		switch (p->last) {
+		case (MD_TEXT):
+			/* FALLTHROUGH */
+		case (MD_OUT):
+			if ( ! mbuf_nputstring(p, " ", 1))
+				return(0);
+			break;
+		default:
+			break;
+		}
+	} else if ( ! mbuf_indent(p))
 		return(0);
+
+	p->last = MD_IN;
+
 	if ( ! mbuf_nputstring(p, "<", 1))
 		return(0);
+	if ( ! mbuf_nputstring(p, "inline:", 7))
+		return(0);
 	if ( ! mbuf_putstring(p, toknames[tok]))
 		return(0);
 
@@ -401,7 +426,6 @@ roffin(void *arg, int tok, int space, int *argc, char **argv)
 		if ( ! mbuf_nputstring(p, "\"", 1))
 			return(0);
 	}
-
 	return(mbuf_nputstring(p, ">", 1));
 }
 
@@ -409,16 +433,20 @@ roffin(void *arg, int tok, int space, int *argc, char **argv)
 static int
 roffout(void *arg, int tok)
 {
-	struct md_valid	*p;
+	struct md_xml	*p;
 
 	assert(arg);
-	p = (struct md_valid *)arg;
+	p = (struct md_xml *)arg;
 
 	if (0 == p->pos && ! mbuf_indent(p))
 		return(0);
 
+	p->last = MD_OUT;
+
 	if ( ! mbuf_nputstring(p, "</", 2))
 		return(0);
+	if ( ! mbuf_nputstring(p, "inline:", 7))
+		return(0);
 	if ( ! mbuf_putstring(p, toknames[tok]))
 		return(0);
 	return(mbuf_nputstring(p, ">", 1));
@@ -430,10 +458,10 @@ roffmsg(void *arg, enum roffmsg lvl,
 		const char *buf, const char *pos, char *msg)
 {
 	char		*level;
-	struct md_valid	*p;
+	struct md_xml	*p;
 
 	assert(arg);
-	p = (struct md_valid *)arg;
+	p = (struct md_xml *)arg;
 
 	switch (lvl) {
 	case (ROFF_WARN):
@@ -462,9 +490,13 @@ roffmsg(void *arg, enum roffmsg lvl,
 static int
 roffdata(void *arg, int space, char *buf)
 {
-	struct md_valid	*p;
+	struct md_xml	*p;
 
 	assert(arg);
-	p = (struct md_valid *)arg;
-	return(mbuf_data(p, space, buf));
+	p = (struct md_xml *)arg;
+	if ( ! mbuf_data(p, space, buf))
+		return(0);
+
+	p->last = MD_TEXT;
+	return(1);
 }