===================================================================
RCS file: /cvs/docbook2mdoc/docbook2mdoc.c,v
retrieving revision 1.10
retrieving revision 1.31
diff -u -p -r1.10 -r1.31
--- docbook2mdoc/docbook2mdoc.c	2014/03/29 10:56:21	1.10
+++ docbook2mdoc/docbook2mdoc.c	2014/04/02 12:27:05	1.31
@@ -1,4 +1,4 @@
-/*	$Id: docbook2mdoc.c,v 1.10 2014/03/29 10:56:21 kristaps Exp $ */
+/*	$Id: docbook2mdoc.c,v 1.31 2014/04/02 12:27:05 kristaps Exp $ */
 /*
  * Copyright (c) 2014 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -26,52 +26,16 @@
 #include <string.h>
 #include <unistd.h>
 
-/*
- * All recognised node types.
- */
-enum	nodeid {
-	NODE_ROOT = 0, /* Must comes first. */
-	/* Alpha-ordered hereafter. */
-	NODE_ARG,
-	NODE_CITEREFENTRY,
-	NODE_CMDSYNOPSIS,
-	NODE_CODE,
-	NODE_COMMAND,
-	NODE_FUNCDEF,
-	NODE_FUNCPROTOTYPE,
-	NODE_FUNCSYNOPSIS,
-	NODE_FUNCSYNOPSISINFO,
-	NODE_FUNCTION,
-	NODE_MANVOLNUM, 
-	NODE_OPTION,
-	NODE_PARA,
-	NODE_PARAMDEF,
-	NODE_PARAMETER,
-	NODE_PROGRAMLISTING,
-	NODE_REFCLASS,
-	NODE_REFDESCRIPTOR,
-	NODE_REFENTRY,
-	NODE_REFENTRYTITLE,
-	NODE_REFMETA,
-	NODE_REFMISCINFO,
-	NODE_REFNAME,
-	NODE_REFNAMEDIV,
-	NODE_REFPURPOSE,
-	NODE_REFSECT1,
-	NODE_REFSYNOPSISDIV,
-	NODE_STRUCTNAME,
-	NODE_SYNOPSIS,
-	NODE_TEXT, 
-	NODE_TITLE,
-	NODE__MAX
-};
+#include "extern.h"
 
 /*
  * Global parse state.
  * Keep this as simple and small as possible.
  */
 struct	parse {
+	XML_Parser	 xml;
 	enum nodeid	 node; /* current (NODE_ROOT if pre-tree) */
+	const char	*fname; /* filename */
 	int		 stop; /* should we stop now? */
 	struct pnode	*root; /* root of parse tree */
 	struct pnode	*cur; /* current node in tree */
@@ -88,310 +52,119 @@ struct	node {
 };
 
 TAILQ_HEAD(pnodeq, pnode);
+TAILQ_HEAD(pattrq, pattr);
 
+struct	pattr {
+	enum attrkey	 key;
+	enum attrval	 val;
+	char		*rawval;
+	TAILQ_ENTRY(pattr) child;
+};
+
 struct	pnode {
 	enum nodeid	 node; /* node type */
 	char		*b; /* binary data buffer */
 	size_t		 bsz; /* data buffer size */
 	struct pnode	*parent; /* parent (or NULL if top) */
 	struct pnodeq	 childq; /* queue of children */
+	struct pattrq	 attrq; /* attributes of node */
 	TAILQ_ENTRY(pnode) child;
 };
 
+static	const char *attrkeys[ATTRKEY__MAX] = {
+	"choice",
+	"id",
+	"rep"
+};
+
+static	const char *attrvals[ATTRVAL__MAX] = {
+	"norepeat",
+	"opt",
+	"plain",
+	"repeat",
+	"req"
+};
+
 static	const struct node nodes[NODE__MAX] = {
 	{ NULL, 0 }, 
+	{ "acronym", 0 }, 
+	{ "anchor", NODE_IGNTEXT }, 
+	{ "application", 0 }, 
 	{ "arg", 0 }, 
+	{ "caution", NODE_IGNTEXT }, 
 	{ "citerefentry", NODE_IGNTEXT }, 
 	{ "cmdsynopsis", NODE_IGNTEXT }, 
 	{ "code", 0 }, 
 	{ "command", 0 }, 
+	{ "date", 0 }, 
+	{ "emphasis", 0 },
+	{ "entry", 0 },
+	{ "envar", 0 },
+	{ "filename", 0 },
 	{ "funcdef", 0 },
 	{ "funcprototype", NODE_IGNTEXT }, 
 	{ "funcsynopsis", NODE_IGNTEXT }, 
 	{ "funcsynopsisinfo", 0 }, 
 	{ "function", 0 }, 
+	{ "itemizedlist", NODE_IGNTEXT }, 
+	{ "group", NODE_IGNTEXT }, 
+	{ "link", 0 }, 
+	{ "listitem", NODE_IGNTEXT }, 
+	{ "literal", 0 }, 
 	{ "manvolnum", 0 }, 
+	{ "note", NODE_IGNTEXT }, 
 	{ "option", 0 }, 
+	{ "orderedlist", NODE_IGNTEXT }, 
 	{ "para", 0 }, 
 	{ "paramdef", 0 },
 	{ "parameter", 0 }, 
 	{ "programlisting", 0 }, 
+	{ "prompt", 0 }, 
+	{ "quote", 0 }, 
 	{ "refclass", NODE_IGNTEXT }, 
 	{ "refdescriptor", NODE_IGNTEXT }, 
 	{ "refentry", NODE_IGNTEXT }, 
+	{ "refentryinfo", NODE_IGNTEXT }, 
 	{ "refentrytitle", 0 }, 
 	{ "refmeta", NODE_IGNTEXT }, 
 	{ "refmiscinfo", NODE_IGNTEXT }, 
 	{ "refname", 0 }, 
 	{ "refnamediv", NODE_IGNTEXT }, 
 	{ "refpurpose", 0 }, 
-	{ "refsect1", 0 }, 
+	{ "refsect1", NODE_IGNTEXT }, 
+	{ "refsect2", NODE_IGNTEXT }, 
+	{ "refsect3", NODE_IGNTEXT }, 
+	{ "refsection", NODE_IGNTEXT }, 
 	{ "refsynopsisdiv", NODE_IGNTEXT }, 
+	{ "replaceable", 0 }, 
+	{ "row", NODE_IGNTEXT }, 
+	{ "sbr", NODE_IGNTEXT }, 
+	{ "screen", NODE_IGNTEXT }, 
+	{ "sgmltag", 0 }, 
 	{ "structname", 0 }, 
 	{ "synopsis", 0 }, 
+	{ "table", NODE_IGNTEXT }, 
+	{ "tbody", NODE_IGNTEXT }, 
+	{ "term", 0 }, 
 	{ NULL, 0 }, 
+	{ "tfoot", NODE_IGNTEXT }, 
+	{ "tgroup", NODE_IGNTEXT }, 
+	{ "thead", NODE_IGNTEXT }, 
+	{ "tip", NODE_IGNTEXT }, 
 	{ "title", 0 }, 
+	{ "ulink", 0 }, 
+	{ "userinput", 0 }, 
+	{ "variablelist", NODE_IGNTEXT }, 
+	{ "varlistentry", NODE_IGNTEXT }, 
+	{ "varname", 0 }, 
+	{ "warning", NODE_IGNTEXT }, 
+	{ "wordasword", 0 }, 
 };
 
 static void
 pnode_print(struct parse *p, struct pnode *pn);
 
 /*
- * Look up whether "parent" is a valid parent for "node".
- * This is sucked directly from the DocBook specification: look at the
- * "children" and "parent" sections of each node.
- */
-static int
-isparent(enum nodeid node, enum nodeid parent)
-{
-
-	switch (node) {
-	case (NODE_ROOT):
-		return(0);
-	case (NODE_ARG):
-		switch (parent) {
-		case (NODE_ARG):
-		case (NODE_CMDSYNOPSIS):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_CITEREFENTRY):
-		switch (parent) {
-		case (NODE_FUNCSYNOPSISINFO):
-		case (NODE_PARA):
-		case (NODE_PROGRAMLISTING):
-		case (NODE_REFDESCRIPTOR):
-		case (NODE_REFENTRYTITLE):
-		case (NODE_REFNAME):
-		case (NODE_REFPURPOSE):
-		case (NODE_SYNOPSIS):
-		case (NODE_TITLE):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_CMDSYNOPSIS):
-		switch (parent) {
-		case (NODE_PARA):
-		case (NODE_REFSECT1):
-		case (NODE_REFSYNOPSISDIV):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_CODE):
-		switch (parent) {
-		case (NODE_FUNCSYNOPSISINFO):
-		case (NODE_PARA):
-		case (NODE_PROGRAMLISTING):
-		case (NODE_REFDESCRIPTOR):
-		case (NODE_REFENTRYTITLE):
-		case (NODE_REFNAME):
-		case (NODE_REFPURPOSE):
-		case (NODE_SYNOPSIS):
-		case (NODE_TITLE):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_COMMAND):
-		switch (parent) {
-		case (NODE_CMDSYNOPSIS):
-		case (NODE_FUNCSYNOPSISINFO):
-		case (NODE_PARA):
-		case (NODE_PROGRAMLISTING):
-		case (NODE_REFDESCRIPTOR):
-		case (NODE_REFENTRYTITLE):
-		case (NODE_REFNAME):
-		case (NODE_REFPURPOSE):
-		case (NODE_SYNOPSIS):
-		case (NODE_TITLE):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_FUNCDEF):
-		return(NODE_FUNCPROTOTYPE == parent);
-	case (NODE_FUNCPROTOTYPE):
-		return(NODE_FUNCSYNOPSIS == parent);
-	case (NODE_FUNCSYNOPSIS):
-		switch (parent) {
-		case (NODE_PARA):
-		case (NODE_REFSECT1):
-		case (NODE_REFSYNOPSISDIV):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_FUNCSYNOPSISINFO):
-		return(NODE_FUNCSYNOPSIS == parent);
-	case (NODE_FUNCTION):
-		switch (parent) {
-		case (NODE_CODE):
-		case (NODE_FUNCDEF):
-		case (NODE_FUNCSYNOPSISINFO):
-		case (NODE_PARA):
-		case (NODE_PROGRAMLISTING):
-		case (NODE_REFDESCRIPTOR):
-		case (NODE_REFENTRYTITLE):
-		case (NODE_REFNAME):
-		case (NODE_REFPURPOSE):
-		case (NODE_SYNOPSIS):
-		case (NODE_TITLE):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_MANVOLNUM):
-		switch (parent) {
-		case (NODE_CITEREFENTRY):
-		case (NODE_REFMETA):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_OPTION):
-		switch (parent) {
-		case (NODE_ARG):
-		case (NODE_FUNCSYNOPSISINFO):
-		case (NODE_PARA):
-		case (NODE_PROGRAMLISTING):
-		case (NODE_REFDESCRIPTOR):
-		case (NODE_REFENTRYTITLE):
-		case (NODE_REFNAME):
-		case (NODE_REFPURPOSE):
-		case (NODE_SYNOPSIS):
-		case (NODE_TITLE):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_PARA):
-		switch (parent) {
-		case (NODE_REFSECT1):
-		case (NODE_REFSYNOPSISDIV):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_PARAMDEF):
-		return(NODE_FUNCPROTOTYPE == parent);
-	case (NODE_PARAMETER):
-		switch (parent) {
-		case (NODE_CODE):
-		case (NODE_FUNCSYNOPSISINFO):
-		case (NODE_PARA):
-		case (NODE_PARAMDEF):
-		case (NODE_PROGRAMLISTING):
-		case (NODE_REFDESCRIPTOR):
-		case (NODE_REFENTRYTITLE):
-		case (NODE_REFNAME):
-		case (NODE_REFPURPOSE):
-		case (NODE_SYNOPSIS):
-		case (NODE_TITLE):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_PROGRAMLISTING):
-		switch (parent) {
-		case (NODE_PARA):
-		case (NODE_REFSECT1):
-		case (NODE_REFSYNOPSISDIV):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_REFCLASS):
-		return(parent == NODE_REFNAMEDIV);
-	case (NODE_REFDESCRIPTOR):
-		return(parent == NODE_REFNAMEDIV);
-	case (NODE_REFENTRY):
-		return(parent == NODE_ROOT);
-	case (NODE_REFENTRYTITLE):
-		switch (parent) {
-		case (NODE_CITEREFENTRY):
-		case (NODE_REFMETA):
-			return(1);
-		default:
-			break;
-		}
-	case (NODE_REFMETA):
-		return(parent == NODE_REFENTRY);
-	case (NODE_REFMISCINFO):
-		return(parent == NODE_REFMETA);
-	case (NODE_REFNAME):
-		return(parent == NODE_REFNAMEDIV);
-	case (NODE_REFNAMEDIV):
-		return(parent == NODE_REFENTRY);
-	case (NODE_REFPURPOSE):
-		return(parent == NODE_REFNAMEDIV);
-	case (NODE_REFSECT1):
-		return(parent == NODE_REFENTRY);
-	case (NODE_REFSYNOPSISDIV):
-		return(parent == NODE_REFENTRY);
-	case (NODE_STRUCTNAME):
-		switch (parent) {
-		case (NODE_CODE):
-		case (NODE_FUNCSYNOPSISINFO):
-		case (NODE_FUNCTION):
-		case (NODE_OPTION):
-		case (NODE_PARA):
-		case (NODE_PARAMETER):
-		case (NODE_PROGRAMLISTING):
-		case (NODE_REFDESCRIPTOR):
-		case (NODE_REFENTRYTITLE):
-		case (NODE_REFNAME):
-		case (NODE_REFPURPOSE):
-		case (NODE_SYNOPSIS):
-		case (NODE_TITLE):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_SYNOPSIS):
-		switch (parent) {
-		case (NODE_REFSYNOPSISDIV):
-		case (NODE_REFSECT1):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_TITLE):
-		switch (parent) {
-		case (NODE_REFSECT1):
-		case (NODE_REFSYNOPSISDIV):
-			return(1);
-		default:
-			break;
-		}
-		return(0);
-	case (NODE_TEXT):
-		return(1);
-	case (NODE__MAX):
-		break;
-	}
-
-	abort();
-	return(0);
-}
-
-/*
  * Process a stream of characters.
  * We store text as nodes in and of themselves.
  * If a text node is already open, append to it.
@@ -437,6 +210,7 @@ xml_char(void *arg, const XML_Char *p, int sz)
 		dat->node = ps->node = NODE_TEXT;
 		dat->parent = ps->cur;
 		TAILQ_INIT(&dat->childq);
+		TAILQ_INIT(&dat->attrq);
 		TAILQ_INSERT_TAIL(&ps->cur->childq, dat, child);
 		ps->cur = dat;
 		assert(NULL != ps->root);
@@ -481,9 +255,13 @@ pnode_trim(struct pnode *pn)
 static void
 xml_elem_start(void *arg, const XML_Char *name, const XML_Char **atts)
 {
-	struct parse	*ps = arg;
-	enum nodeid	 node;
-	struct pnode	*dat;
+	struct parse	 *ps = arg;
+	enum nodeid	  node;
+	enum attrkey	  key;
+	enum attrval	  val;
+	struct pnode	 *dat;
+	struct pattr	 *pattr;
+	const XML_Char	**att;
 
 	if (ps->stop)
 		return;
@@ -503,23 +281,31 @@ xml_elem_start(void *arg, const XML_Char *name, const 
 		else if (0 == strcmp(nodes[node].name, name))
 			break;
 
-	/* FIXME: do more with these error messages... */
 	if (NODE__MAX == node && NODE_ROOT == ps->node) {
-		fprintf(stderr, "%s: ignoring node\n", name);
 		return;
 	} else if (NODE__MAX == node) {
-		fprintf(stderr, "%s: unknown node\n", name);
+		fprintf(stderr, "%s:%zu:%zu: unknown node \"%s\"\n", 
+			ps->fname, XML_GetCurrentLineNumber(ps->xml),
+			XML_GetCurrentColumnNumber(ps->xml), name);
 		ps->stop = 1;
 		return;
 	} else if (NODE_ROOT == ps->node && NULL != ps->root) {
-		fprintf(stderr, "%s: reentering?\n", name);
+		fprintf(stderr, "%s:%zu:%zu: multiple refentries\n",
+			ps->fname, XML_GetCurrentLineNumber(ps->xml),
+			XML_GetCurrentColumnNumber(ps->xml));
 		ps->stop = 1;
 		return;
 	} else if (NODE_ROOT == ps->node && NODE_REFENTRY != node) {
-		fprintf(stderr, "%s: known node w/o context\n", name);
 		return;
 	} else if ( ! isparent(node, ps->node)) {
-		fprintf(stderr, "%s: bad parent\n", name);
+		fprintf(stderr, "%s:%zu:%zu: bad parent \"%s\" "
+			"of node \"%s\"\n",
+			ps->fname, XML_GetCurrentLineNumber(ps->xml),
+			XML_GetCurrentColumnNumber(ps->xml),
+			NULL == nodes[ps->node].name ?
+			"(none)" : nodes[ps->node].name,
+			NULL == nodes[node].name ?
+			"(none)" : nodes[node].name);
 		ps->stop = 1;
 		return;
 	}
@@ -532,6 +318,7 @@ xml_elem_start(void *arg, const XML_Char *name, const 
 	dat->node = ps->node = node;
 	dat->parent = ps->cur;
 	TAILQ_INIT(&dat->childq);
+	TAILQ_INIT(&dat->attrq);
 
 	if (NULL != ps->cur)
 		TAILQ_INSERT_TAIL(&ps->cur->childq, dat, child);
@@ -539,6 +326,51 @@ xml_elem_start(void *arg, const XML_Char *name, const 
 	ps->cur = dat;
 	if (NULL == ps->root)
 		ps->root = dat;
+
+	/*
+	 * Process attributes.
+	 */
+	for (att = atts; NULL != *att; att += 2) {
+		for (key = 0; key < ATTRKEY__MAX; key++)
+			if (0 == strcmp(*att, attrkeys[key]))
+				break;
+		if (ATTRKEY__MAX == key) {
+			fprintf(stderr, "%s:%zu:%zu: warning: "
+				"unknown attribute \"%s\"\n", 
+				ps->fname, 
+				XML_GetCurrentLineNumber(ps->xml),
+				XML_GetCurrentColumnNumber(ps->xml), 
+				*att);
+			continue;
+		} else if ( ! isattrkey(node, key)) {
+			fprintf(stderr, "%s:%zu:%zu: warning: "
+				"bad attribute \"%s\"\n", 
+				ps->fname, 
+				XML_GetCurrentLineNumber(ps->xml),
+				XML_GetCurrentColumnNumber(ps->xml), 
+				*att);
+			continue;
+		}
+		for (val = 0; val < ATTRVAL__MAX; val++)
+			if (0 == strcmp(*(att + 1), attrvals[val]))
+				break;
+		if (ATTRVAL__MAX != val && ! isattrval(key, val)) {
+			fprintf(stderr, "%s:%zu:%zu: warning: "
+				"bad attribute value \"%s\"\n", 
+				ps->fname, 
+				XML_GetCurrentLineNumber(ps->xml),
+				XML_GetCurrentColumnNumber(ps->xml), 
+				*(att + 1));
+			continue;
+		}
+		pattr = calloc(1, sizeof(struct pattr));
+		pattr->key = key;
+		pattr->val = val;
+		if (ATTRVAL__MAX == val)
+			pattr->rawval = strdup(*(att + 1));
+		TAILQ_INSERT_TAIL(&dat->attrq, pattr, child);
+	}
+
 }
 
 /*
@@ -576,6 +408,7 @@ static void
 pnode_free(struct pnode *pn)
 {
 	struct pnode	*pp;
+	struct pattr	*ap;
 
 	if (NULL == pn)
 		return;
@@ -585,6 +418,12 @@ pnode_free(struct pnode *pn)
 		pnode_free(pp);
 	}
 
+	while (NULL != (ap = TAILQ_FIRST(&pn->attrq))) {
+		TAILQ_REMOVE(&pn->attrq, ap, child);
+		free(ap->rawval);
+		free(ap);
+	}
+
 	free(pn->b);
 	free(pn);
 }
@@ -661,15 +500,39 @@ bufappend_r(struct parse *p, struct pnode *pn)
 		bufappend_r(p, pp);
 }
 
+/* 
+ * Recursively search and return the first instance of "node".
+ */
+static struct pnode *
+pnode_findfirst(struct pnode *pn, enum nodeid node)
+{
+	struct pnode	*pp, *res;
+
+	res = NULL;
+	TAILQ_FOREACH(pp, &pn->childq, child) {
+		res = pp->node == node ? pp :
+			pnode_findfirst(pp, node);
+		if (NULL != res)
+			break;
+	}
+
+	return(res);
+}
+
+#define MACROLINE_NORM	0
+#define MACROLINE_UPPER	1
 /*
  * Recursively print text presumably on a macro line.
  * Convert all whitespace to regular spaces.
  */
 static void
-pnode_printmacrolinepart(struct parse *p, struct pnode *pn)
+pnode_printmacrolinetext(struct parse *p, struct pnode *pn, int fl)
 {
 	char		*cp;
 
+	if (0 == p->newln)
+		putchar(' ');
+
 	bufclear(p);
 	bufappend_r(p, pn);
 
@@ -692,13 +555,23 @@ pnode_printmacrolinepart(struct parse *p, struct pnode
 			  ('\0' == *(cp + 3) || 
 			   ' ' == *(cp + 3)))))
 			fputs("\\&", stdout);
-		putchar(*cp);
+		if (MACROLINE_UPPER & fl)
+			putchar(toupper((int)*cp));
+		else
+			putchar((int)*cp);
 		/* If we're a character escape, escape us. */
 		if ('\\' == *cp)
 			putchar('e');
 	}
 }
 
+static void
+pnode_printmacrolinepart(struct parse *p, struct pnode *pn)
+{
+
+	pnode_printmacrolinetext(p, pn, 0);
+}
+
 /*
  * Just pnode_printmacrolinepart() but with a newline.
  * If no text, just the newline.
@@ -707,8 +580,10 @@ static void
 pnode_printmacroline(struct parse *p, struct pnode *pn)
 {
 
-	pnode_printmacrolinepart(p, pn);
+	assert(0 == p->newln);
+	pnode_printmacrolinetext(p, pn, 0);
 	putchar('\n');
+	p->newln = 1;
 }
 
 static void
@@ -758,13 +633,48 @@ pnode_printrefsect(struct parse *p, struct pnode *pn)
 		if (NODE_TITLE == pp->node)
 			break;
 
-	fputs(".Sh ", stdout);
+	switch (pn->node) {
+	case (NODE_REFSECT1):
+		fputs(".Sh", stdout);
+		break;
+	case (NODE_REFSECT2):
+		fputs(".Ss", stdout);
+		break;
+	case (NODE_REFSECT3):
+		puts(".Pp");
+		fputs(".Sy", stdout);
+		break;
+	case (NODE_NOTE):
+		/* FALLTHROUGH */
+	case (NODE_REFSECTION):
+		/* FALLTHROUGH */
+	case (NODE_TIP):
+		/* FALLTHROUGH */
+	case (NODE_CAUTION):
+		/* FALLTHROUGH */
+	case (NODE_WARNING):
+		puts(".Pp");
+		if (NULL == pp)
+			return;
+		fputs(".Em", stdout);
+		break;
+	default:
+		break;
+	}
 
+	p->newln = 0;
+
 	if (NULL != pp) {
-		pnode_printmacroline(p, pp);
+		pnode_printmacrolinetext(p, pp, 
+			NODE_REFSECT1 == pn->node ?
+			MACROLINE_UPPER : 0);
+		pnode_printmclose(p, 1);
 		pnode_unlink(pp);
-	} else
-		puts("UNKNOWN");
+	} else {
+		puts(NODE_REFSECT1 == pn->node ?
+			"UNKNOWN" : "unknown");
+		p->newln = 1;
+	}
 }
 
 /*
@@ -776,24 +686,26 @@ pnode_printciterefentry(struct parse *p, struct pnode 
 	struct pnode	*pp, *title, *manvol;
 
 	title = manvol = NULL;
+	assert(p->newln);
 	TAILQ_FOREACH(pp, &pn->childq, child)
 		if (NODE_MANVOLNUM == pp->node)
 			manvol = pp;
 		else if (NODE_REFENTRYTITLE == pp->node)
 			title = pp;
 
-	fputs(".Xr ", stdout);
+	fputs(".Xr", stdout);
+	p->newln = 0;
 
 	if (NULL != title) {
 		pnode_printmacrolinepart(p, title);
-		putchar(' ');
 	} else
-		fputs("unknown ", stdout);
+		fputs(" unknown ", stdout);
 
-	if (NULL != manvol)
+	if (NULL == manvol) {
+		puts(" 1");
+		p->newln = 1;
+	} else
 		pnode_printmacroline(p, manvol);
-	else
-		puts("1");
 }
 
 static void
@@ -802,6 +714,7 @@ pnode_printrefmeta(struct parse *p, struct pnode *pn)
 	struct pnode	*pp, *title, *manvol;
 
 	title = manvol = NULL;
+	assert(p->newln);
 	TAILQ_FOREACH(pp, &pn->childq, child)
 		if (NODE_MANVOLNUM == pp->node)
 			manvol = pp;
@@ -809,19 +722,19 @@ pnode_printrefmeta(struct parse *p, struct pnode *pn)
 			title = pp;
 
 	puts(".Dd $Mdocdate" "$");
-	fputs(".Dt ", stdout);
+	fputs(".Dt", stdout);
+	p->newln = 0;
 
-	if (NULL != title) {
-		/* FIXME: uppercase. */
-		pnode_printmacrolinepart(p, title);
-		putchar(' ');
-	} else
-		fputs("UNKNOWN ", stdout);
+	if (NULL != title)
+		pnode_printmacrolinetext(p, title, MACROLINE_UPPER);
+	else
+		fputs(" UNKNOWN ", stdout);
 
-	if (NULL != manvol)
+	if (NULL == manvol) {
+		puts(" 1");
+		p->newln = 1;
+	} else
 		pnode_printmacroline(p, manvol);
-	else
-		puts("1");
 
 	puts(".Os");
 }
@@ -831,6 +744,7 @@ pnode_printfuncdef(struct parse *p, struct pnode *pn)
 {
 	struct pnode	*pp, *ftype, *func;
 
+	assert(p->newln);
 	ftype = func = NULL;
 	TAILQ_FOREACH(pp, &pn->childq, child)
 		if (NODE_TEXT == pp->node)
@@ -839,15 +753,19 @@ pnode_printfuncdef(struct parse *p, struct pnode *pn)
 			func = pp;
 
 	if (NULL != ftype) {
-		fputs(".Ft ", stdout);
+		fputs(".Ft", stdout);
+		p->newln = 0;
 		pnode_printmacroline(p, ftype);
 	}
 
 	if (NULL != func) {
-		fputs(".Fo ", stdout);
+		fputs(".Fo", stdout);
+		p->newln = 0;
 		pnode_printmacroline(p, func);
-	} else 
+	} else {
 		puts(".Fo UNKNOWN");
+		p->newln = 1;
+	}
 }
 
 static void
@@ -855,6 +773,7 @@ pnode_printparamdef(struct parse *p, struct pnode *pn)
 {
 	struct pnode	*pp, *ptype, *param;
 
+	assert(p->newln);
 	ptype = param = NULL;
 	TAILQ_FOREACH(pp, &pn->childq, child)
 		if (NODE_TEXT == pp->node)
@@ -863,6 +782,7 @@ pnode_printparamdef(struct parse *p, struct pnode *pn)
 			param = pp;
 
 	fputs(".Fa \"", stdout);
+	p->newln = 0;
 	if (NULL != ptype) {
 		pnode_printmacrolinepart(p, ptype);
 		putchar(' ');
@@ -870,10 +790,9 @@ pnode_printparamdef(struct parse *p, struct pnode *pn)
 
 	if (NULL != param)
 		pnode_printmacrolinepart(p, param);
-	else
-		fputs("UNKNOWN", stdout);
 
 	puts("\"");
+	p->newln = 1;
 }
 
 static void
@@ -881,6 +800,7 @@ pnode_printfuncprototype(struct parse *p, struct pnode
 {
 	struct pnode	*pp, *fdef;
 
+	assert(p->newln);
 	TAILQ_FOREACH(fdef, &pn->childq, child)
 		if (NODE_FUNCDEF == fdef->node) 
 			break;
@@ -895,45 +815,100 @@ pnode_printfuncprototype(struct parse *p, struct pnode
 			pnode_printparamdef(p, pp);
 
 	puts(".Fc");
+	p->newln = 1;
 }
 
 /* 
  * The <arg> element is more complicated than it should be because text
  * nodes are treated like ".Ar foo", but non-text nodes need to be
  * re-sent into the printer (i.e., without the preceding ".Ar").
- * TODO: handle "optional" attribute.
+ * This also handles the case of "repetition" (or in other words, the
+ * ellipsis following an argument) and optionality.
  */
 static void
 pnode_printarg(struct parse *p, struct pnode *pn)
 {
 	struct pnode	*pp;
+	struct pattr	*ap;
+	int		 isop, isrep;
 
+	isop = 1;
+	isrep = 0;
+	TAILQ_FOREACH(ap, &pn->attrq, child) 
+		if (ATTRKEY_CHOICE == ap->key &&
+			(ATTRVAL_PLAIN == ap->val ||
+			 ATTRVAL_REQ == ap->val)) 
+			isop = 0;
+		else if (ATTRKEY_REP == ap->key &&
+			(ATTRVAL_REPEAT == ap->val))
+			isrep = 1;
+
+	if (isop) {
+		pnode_printmopen(p);
+		fputs("Op", stdout);
+	}
+
 	TAILQ_FOREACH(pp, &pn->childq, child) {
 		if (NODE_TEXT == pp->node) {
 			pnode_printmopen(p);
-			fputs("Ar ", stdout);
+			fputs("Ar", stdout);
 		} 
 		pnode_print(p, pp);
+		if (NODE_TEXT == pp->node && isrep) 
+			fputs("...", stdout);
 	}
 }
 
-/* 
- * Recursively search and return the first instance of "node".
- */
-static struct pnode *
-pnode_findfirst(struct pnode *pn, enum nodeid node)
+static void
+pnode_printgroup(struct parse *p, struct pnode *pn)
 {
-	struct pnode	*pp, *res;
+	struct pnode	*pp, *np;
+	struct pattr	*ap;
+	int		 isop, sv;
 
-	res = NULL;
-	TAILQ_FOREACH(pp, &pn->childq, child) {
-		res = pp->node == node ? pp :
-			pnode_findfirst(pp, node);
-		if (NULL != res)
+	isop = 1;
+	TAILQ_FOREACH(ap, &pn->attrq, child) 
+		if (ATTRKEY_CHOICE == ap->key &&
+			(ATTRVAL_PLAIN == ap->val ||
+			 ATTRVAL_REQ == ap->val)) {
+			isop = 0;
 			break;
+		}
+
+	/* 
+	 * Make sure we're on a macro line.
+	 * This will prevent pnode_print() for putting us on a
+	 * subsequent line.
+	 */
+	sv = p->newln;
+	pnode_printmopen(p);
+	if (isop) 
+		fputs("Op", stdout);
+	else if (sv)
+		fputs("No", stdout);
+
+	/*
+	 * Keep on printing text separated by the vertical bar as long
+	 * as we're within the same origin node as the group.
+	 * This is kind of a nightmare.
+	 * Eh, DocBook...
+	 * FIXME: if there's a "Fl", we don't cut off the leading "-"
+	 * like we do in pnode_print().
+	 */
+	TAILQ_FOREACH(pp, &pn->childq, child) {
+		pnode_print(p, pp);
+		np = TAILQ_NEXT(pp, child);
+		while (NULL != np) {
+			if (pp->node != np->node)
+				break;
+			fputs(" |", stdout);
+			pnode_printmacrolinepart(p, np);
+			pp = np;
+			np = TAILQ_NEXT(np, child);
+		}
 	}
 
-	return(res);
+	pnode_printmclose(p, sv);
 }
 
 static void
@@ -955,10 +930,129 @@ pnode_printprologue(struct parse *p, struct pnode *pn)
 	}
 }
 
+static void
+pnode_printvarlistentry(struct parse *p, struct pnode *pn)
+{
+	struct pnode	*pp;
+
+	assert(p->newln);
+	TAILQ_FOREACH(pp, &pn->childq, child)
+		if (NODE_TERM == pp->node) {
+			fputs(".It", stdout);
+			p->newln = 0;
+			pnode_print(p, pp);
+			pnode_unlink(pp);
+			pnode_printmclose(p, 1);
+			return;
+		} 
+
+	puts(".It");
+	p->newln = 1;
+}
+
+static void
+pnode_printrow(struct parse *p, struct pnode *pn)
+{
+	struct pnode	*pp;
+
+	puts(".Bl -dash -compact");
+
+	TAILQ_FOREACH(pp, &pn->childq, child) {
+		assert(p->newln);
+		puts(".It");
+		pnode_print(p, pp);
+		pnode_printmclose(p, 1);
+	}
+	assert(p->newln);
+	puts(".El");
+}
+
+static void
+pnode_printtable(struct parse *p, struct pnode *pn)
+{
+	struct pnode	*pp;
+
+	assert(p->newln);
+	TAILQ_FOREACH(pp, &pn->childq, child)
+		if (NODE_TITLE == pp->node) {
+			puts(".Pp");
+			pnode_print(p, pp);
+			pnode_unlink(pp);
+		}
+	assert(p->newln);
+	puts(".Bl -ohang");
+	while (NULL != (pp = pnode_findfirst(pn, NODE_ROW))) {
+		puts(".It Table Row");
+		pnode_printrow(p, pp);
+		pnode_printmclose(p, 1);
+		pnode_unlink(pp);
+	}
+	assert(p->newln);
+	puts(".El");
+}
+
+static void
+pnode_printlist(struct parse *p, struct pnode *pn)
+{
+	struct pnode	*pp;
+
+	assert(p->newln);
+	TAILQ_FOREACH(pp, &pn->childq, child)
+		if (NODE_TITLE == pp->node) {
+			puts(".Pp");
+			pnode_print(p, pp);
+			pnode_unlink(pp);
+		}
+	assert(p->newln);
+
+	if (NODE_ORDEREDLIST == pn->node)
+		puts(".Bl -enum");
+	else
+		puts(".Bl -item");
+
+	TAILQ_FOREACH(pp, &pn->childq, child) {
+		assert(p->newln);
+		puts(".It");
+		pnode_print(p, pp);
+		pnode_printmclose(p, 1);
+	}
+	assert(p->newln);
+	puts(".El");
+}
+
+static void
+pnode_printvariablelist(struct parse *p, struct pnode *pn)
+{
+	struct pnode	*pp;
+
+	assert(p->newln);
+	TAILQ_FOREACH(pp, &pn->childq, child)
+		if (NODE_TITLE == pp->node) {
+			puts(".Pp");
+			pnode_print(p, pp);
+			pnode_unlink(pp);
+		}
+
+	assert(p->newln);
+	puts(".Bl -tag -width Ds");
+	TAILQ_FOREACH(pp, &pn->childq, child)
+		if (NODE_VARLISTENTRY != pp->node) {
+			assert(p->newln);
+			fputs(".It", stdout);
+			pnode_printmacroline(p, pp);
+		} else {
+			assert(p->newln);
+			pnode_print(p, pp);
+		}
+	assert(p->newln);
+	puts(".El");
+}
+
 /*
  * Print a parsed node (or ignore it--whatever).
  * This is a recursive function.
- * FIXME: macro line continuation?
+ * FIXME: if we're in a literal context (<screen> or <programlisting> or
+ * whatever), don't print inline macros.
  */
 static void
 pnode_print(struct parse *p, struct pnode *pn)
@@ -973,6 +1067,13 @@ pnode_print(struct parse *p, struct pnode *pn)
 	sv = p->newln;
 
 	switch (pn->node) {
+	case (NODE_APPLICATION):
+		pnode_printmopen(p);
+		fputs("Nm", stdout);
+		break;
+	case (NODE_ANCHOR):
+		/* Don't print anything! */
+		return;
 	case (NODE_ARG):
 		pnode_printarg(p, pn);
 		pnode_unlinksub(pn);
@@ -984,15 +1085,27 @@ pnode_print(struct parse *p, struct pnode *pn)
 		break;
 	case (NODE_CODE):
 		pnode_printmopen(p);
-		fputs("Li ", stdout);
+		fputs("Li", stdout);
 		break;
 	case (NODE_COMMAND):
 		pnode_printmopen(p);
-		fputs("Nm ", stdout);
+		fputs("Nm", stdout);
 		break;
+	case (NODE_EMPHASIS):
+		pnode_printmopen(p);
+		fputs("Em", stdout);
+		break;
+	case (NODE_ENVAR):
+		pnode_printmopen(p);
+		fputs("Ev", stdout);
+		break;
+	case (NODE_FILENAME):
+		pnode_printmopen(p);
+		fputs("Pa", stdout);
+		break;
 	case (NODE_FUNCTION):
 		pnode_printmopen(p);
-		fputs("Fn ", stdout);
+		fputs("Fn", stdout);
 		break;
 	case (NODE_FUNCPROTOTYPE):
 		assert(p->newln);
@@ -1001,14 +1114,35 @@ pnode_print(struct parse *p, struct pnode *pn)
 		break;
 	case (NODE_FUNCSYNOPSISINFO):
 		pnode_printmopen(p);
-		fputs("Fd ", stdout);
+		fputs("Fd", stdout);
 		break;
+	case (NODE_ITEMIZEDLIST):
+		assert(p->newln);
+		pnode_printlist(p, pn);
+		pnode_unlinksub(pn);
+		break;
+	case (NODE_GROUP):
+		pnode_printgroup(p, pn);
+		pnode_unlinksub(pn);
+		break;
+	case (NODE_LITERAL):
+		pnode_printmopen(p);
+		fputs("Li", stdout);
+		break;
 	case (NODE_OPTION):
 		pnode_printmopen(p);
-		fputs("Fl ", stdout);
+		fputs("Fl", stdout);
 		break;
+	case (NODE_ORDEREDLIST):
+		assert(p->newln);
+		pnode_printlist(p, pn);
+		pnode_unlinksub(pn);
+		break;
 	case (NODE_PARA):
 		assert(p->newln);
+		if (NULL != pn->parent && 
+			NODE_LISTITEM == pn->parent->node) 
+			break;
 		puts(".Pp");
 		break;
 	case (NODE_PARAMETER):
@@ -1019,17 +1153,28 @@ pnode_print(struct parse *p, struct pnode *pn)
 		puts("\"");
 		pnode_unlinksub(pn);
 		break;
+	case (NODE_QUOTE):
+		pnode_printmopen(p);
+		fputs("Qo", stdout);
+		break;
 	case (NODE_PROGRAMLISTING):
+		/* FALLTHROUGH */
+	case (NODE_SCREEN):
 		assert(p->newln);
 		puts(".Bd -literal");
 		break;
+	case (NODE_REFENTRYINFO):
+		/* Suppress. */
+		pnode_unlinksub(pn);
+		break;
 	case (NODE_REFMETA):
 		abort();
 		break;
 	case (NODE_REFNAME):
 		/* Suppress non-text children... */
 		pnode_printmopen(p);
-		fputs("Nm ", stdout);
+		fputs("Nm", stdout);
+		p->newln = 0;
 		pnode_printmacrolinepart(p, pn);
 		pnode_unlinksub(pn);
 		break;
@@ -1039,7 +1184,8 @@ pnode_print(struct parse *p, struct pnode *pn)
 		break;
 	case (NODE_REFPURPOSE):
 		assert(p->newln);
-		fputs(".Nd ", stdout);
+		pnode_printmopen(p);
+		fputs("Nd", stdout);
 		break;
 	case (NODE_REFSYNOPSISDIV):
 		assert(p->newln);
@@ -1047,14 +1193,47 @@ pnode_print(struct parse *p, struct pnode *pn)
 		puts(".Sh SYNOPSIS");
 		break;
 	case (NODE_REFSECT1):
+		/* FALLTHROUGH */
+	case (NODE_REFSECT2):
+		/* FALLTHROUGH */
+	case (NODE_REFSECT3):
+		/* FALLTHROUGH */
+	case (NODE_REFSECTION):
+		/* FALLTHROUGH */
+	case (NODE_NOTE):
+		/* FALLTHROUGH */
+	case (NODE_TIP):
+		/* FALLTHROUGH */
+	case (NODE_CAUTION):
+		/* FALLTHROUGH */
+	case (NODE_WARNING):
 		assert(p->newln);
 		pnode_printrefsect(p, pn);
 		break;
+	case (NODE_REPLACEABLE):
+		pnode_printmopen(p);
+		fputs("Ar", stdout);
+		break;
+	case (NODE_SBR):
+		assert(p->newln);
+		puts(".br");
+		break;
+	case (NODE_SGMLTAG):
+		pnode_printmopen(p);
+		fputs("Li", stdout);
+		break;
 	case (NODE_STRUCTNAME):
 		pnode_printmopen(p);
-		fputs("Vt ", stdout);
+		fputs("Vt", stdout);
 		break;
+	case (NODE_TABLE):
+		assert(p->newln);
+		pnode_printtable(p, pn);
+		pnode_unlinksub(pn);
+		break;
 	case (NODE_TEXT):
+		if (0 == p->newln)
+			putchar(' ');
 		bufclear(p);
 		bufappend(p, pn);
 		/*
@@ -1064,7 +1243,17 @@ pnode_print(struct parse *p, struct pnode *pn)
 		 * Remember to escape control characters and escapes.
 		 */
 		assert(p->bsz);
-		for (last = '\n', cp = p->b; '\0' != *cp; ) {
+		cp = p->b;
+		/*
+		 * There's often a superfluous "-" in its <option> tags
+		 * before the actual flags themselves.
+		 * "Fl" does this for us, so remove it.
+		 */
+		if (NULL != pn->parent &&
+			NODE_OPTION == pn->parent->node &&
+			'-' == *cp)
+			cp++;
+		for (last = '\n'; '\0' != *cp; ) {
 			if ('\n' == last) {
 				/* Consume all whitespace. */
 				if (isspace((int)*cp)) {
@@ -1081,6 +1270,23 @@ pnode_print(struct parse *p, struct pnode *pn)
 		}
 		p->newln = 0;
 		break;
+	case (NODE_USERINPUT):
+		pnode_printmopen(p);
+		fputs("Li", stdout);
+		break;
+	case (NODE_VARIABLELIST):
+		assert(p->newln);
+		pnode_printvariablelist(p, pn);
+		pnode_unlinksub(pn);
+		break;
+	case (NODE_VARLISTENTRY):
+		assert(p->newln);
+		pnode_printvarlistentry(p, pn);
+		break;
+	case (NODE_VARNAME):
+		pnode_printmopen(p);
+		fputs("Va", stdout);
+		break;
 	default:
 		break;
 	}
@@ -1089,19 +1295,50 @@ pnode_print(struct parse *p, struct pnode *pn)
 		pnode_print(p, pp);
 
 	switch (pn->node) {
+	case (NODE_APPLICATION):
 	case (NODE_ARG):
 	case (NODE_CODE):
 	case (NODE_COMMAND):
+	case (NODE_EMPHASIS):
+	case (NODE_ENVAR):
+	case (NODE_FILENAME):
 	case (NODE_FUNCTION):
 	case (NODE_FUNCSYNOPSISINFO):
+	case (NODE_LITERAL):
 	case (NODE_OPTION):
 	case (NODE_PARAMETER):
-	case (NODE_REFNAME):
+	case (NODE_REPLACEABLE):
+	case (NODE_REFPURPOSE):
+	case (NODE_SGMLTAG):
 	case (NODE_STRUCTNAME):
 	case (NODE_TEXT):
+	case (NODE_USERINPUT):
+	case (NODE_VARNAME):
 		pnode_printmclose(p, sv);
 		break;
+	case (NODE_QUOTE):
+		pnode_printmclose(p, sv);
+		sv = p->newln;
+		pnode_printmopen(p);
+		fputs("Qc", stdout);
+		pnode_printmclose(p, sv);
+		break;
+	case (NODE_REFNAME):
+		/*
+		 * If we're in the NAME macro and we have multiple
+		 * <refname> macros in sequence, then print out a
+		 * trailing comma before the newline.
+		 */
+		if (NULL != pn->parent && 
+			NODE_REFNAMEDIV == pn->parent->node &&
+			NULL != TAILQ_NEXT(pn, child) &&
+			NODE_REFNAME == TAILQ_NEXT(pn, child)->node) 
+			fputs(" ,", stdout);
+		pnode_printmclose(p, sv);
+		break;
 	case (NODE_PROGRAMLISTING):
+		/* FALLTHROUGH */
+	case (NODE_SCREEN):
 		assert(p->newln);
 		puts(".Ed");
 		p->newln = 1;
@@ -1126,6 +1363,8 @@ readfile(XML_Parser xp, int fd, 
 	memset(&p, 0, sizeof(struct parse));
 
 	p.b = malloc(p.bsz = p.mbsz = 1024);
+	p.fname = fn;
+	p.xml = xp;
 
 	XML_SetCharacterDataHandler(xp, xml_char);
 	XML_SetElementHandler(xp, xml_elem_start, xml_elem_end);
@@ -1133,7 +1372,9 @@ readfile(XML_Parser xp, int fd, 
 
 	while ((ssz = read(fd, b, bsz)) >= 0) {
 		if (0 == (rc = XML_Parse(xp, b, ssz, 0 == ssz)))
-			fprintf(stderr, "%s: %s\n", fn,
+			fprintf(stderr, "%s:%zu:%zu: %s\n", fn,
+				XML_GetCurrentLineNumber(xp),
+				XML_GetCurrentColumnNumber(xp), 
 				XML_ErrorString
 				(XML_GetErrorCode(xp)));
 		else if ( ! p.stop && ssz > 0)