Sync to version 1.12.0; all code by kristaps@:

Implement .Rv in -Tman.
Let -man -Tman work a bit like cat(1).
Add the -Ofragment option to -T[x]html.
Minor fixes in -T[x]html.
Lots of apropos(1) and -Tman code cleanup.

17 files changed, 517 insertions, 348 deletions

diff --git a/usr.bin/mandoc/Makefile b/usr.bin/mandoc/Makefile
index 0f93eed462b..ca988968211 100644
--- a/usr.bin/mandoc/Makefile
+++ b/usr.bin/mandoc/Makefile
@@ -1,8 +1,8 @@
-#	$OpenBSD: Makefile,v 1.61 2011/10/06 23:04:16 schwarze Exp $
+#	$OpenBSD: Makefile,v 1.62 2011/10/09 17:59:56 schwarze Exp $
 
 .include <bsd.own.mk>
 
-CFLAGS+=-DVERSION=\"1.11.7\"
+CFLAGS+=-DVERSION=\"1.12.0\"
 CFLAGS+=-W -Wall -Wstrict-prototypes
 
 .if ${COMPILER_VERSION:L} == "gcc3" || ${COMPILER_VERSION:L} == "gcc4" 
diff --git a/usr.bin/mandoc/apropos.1 b/usr.bin/mandoc/apropos.1
index b5fda7ab22c..281e2ececf0 100644
--- a/usr.bin/mandoc/apropos.1
+++ b/usr.bin/mandoc/apropos.1
@@ -1,4 +1,4 @@
-.\"	$Id: apropos.1,v 1.1 2011/10/06 23:04:16 schwarze Exp $
+.\"	$Id: apropos.1,v 1.2 2011/10/09 17:59:56 schwarze Exp $
 .\"
 .\" Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
 .\"
@@ -14,7 +14,7 @@
 .\" ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 .\" OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 .\"
-.Dd $Mdocdate: October 6 2011 $
+.Dd $Mdocdate: October 9 2011 $
 .Dt APROPOS 1
 .Os
 .Sh NAME
@@ -162,7 +162,8 @@ in any letter case:
 The
 .Nm
 utility was written by
-.An Kristaps Dzonsons Aq kristaps@bsd.lv .
+.An Kristaps Dzonsons ,
+.Mt kristaps@bsd.lv .
 .\" .Sh CAVEATS
 .\" .Sh BUGS
 .\" .Sh SECURITY CONSIDERATIONS
diff --git a/usr.bin/mandoc/apropos.c b/usr.bin/mandoc/apropos.c
index 002078010e3..7e2f5722a91 100644
--- a/usr.bin/mandoc/apropos.c
+++ b/usr.bin/mandoc/apropos.c
@@ -1,4 +1,4 @@
-/*	$Id: apropos.c,v 1.1 2011/10/06 23:04:16 schwarze Exp $ */
+/*	$Id: apropos.c,v 1.2 2011/10/09 17:59:56 schwarze Exp $ */
 /*
 * Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -32,21 +32,23 @@
 
 #include "mandoc.h"
 
-#define	MAXRESULTS	 100
-
-#define TYPE_NAME	0x01
-#define TYPE_FUNCTION	0x02
-#define TYPE_UTILITY	0x04
-#define TYPE_INCLUDES	0x08
-#define TYPE_VARIABLE	0x10
-#define TYPE_STANDARD	0x20
-#define TYPE_AUTHOR	0x40
-#define TYPE_CONFIG	0x80
-#define TYPE_DESC	0x100
-#define TYPE_XREF	0x200
-#define TYPE_PATH	0x400
-#define	TYPE_ENV	0x800
-#define	TYPE_ERR	0x1000
+#define	MAXRESULTS	 256
+
+/* Bit-fields.  See mandocdb.8. */
+
+#define TYPE_NAME	  0x01
+#define TYPE_FUNCTION	  0x02
+#define TYPE_UTILITY	  0x04
+#define TYPE_INCLUDES	  0x08
+#define TYPE_VARIABLE	  0x10
+#define TYPE_STANDARD	  0x20
+#define TYPE_AUTHOR	  0x40
+#define TYPE_CONFIG	  0x80
+#define TYPE_DESC	  0x100
+#define TYPE_XREF	  0x200
+#define TYPE_PATH	  0x400
+#define TYPE_ENV	  0x800
+#define TYPE_ERR	  0x1000
 
 enum	match {
 	MATCH_SUBSTR = 0,
@@ -71,16 +73,16 @@ struct	opts {
 
 struct	type {
 	int		 mask;
-	const char	*name;
+	const char	*name; /* command-line type name */
 };
 
 struct	rec {
-	char		*file;
-	char		*cat;
-	char		*title;
-	char		*arch;
-	char		*desc;
-	recno_t		 rec;
+	char		*file; /* file in file-system */
+	char		*cat; /* category (3p, 3, etc.) */
+	char		*title; /* title (FOO, etc.) */
+	char		*arch; /* arch (or empty string) */
+	char		*desc; /* description (from Nd) */
+	recno_t		 rec; /* record in index */
 };
 
 struct	res {
@@ -92,6 +94,14 @@ struct	res {
 	char		*title; /* manual section */
 	char		*uri; /* formatted uri of file */
 	recno_t		 rec; /* unique id of underlying manual */
+	/*
+	 * Maintain a binary tree for checking the uniqueness of `rec'
+	 * when adding elements to the results array.
+	 * Since the results array is dynamic, use offset in the array
+	 * instead of a pointer to the structure.
+	 */
+	int		 lhs;
+	int		 rhs;
 };
 
 struct	state {
@@ -99,8 +109,6 @@ struct	state {
 	DB		 *idx; /* index */
 	const char	 *dbf; /* database name */
 	const char	 *idxf; /* index name */
-	void		(*err)(const char *);
-	void		(*errx)(const char *, ...);
 };
 
 static	const char * const sorts[SORT__MAX] = {
@@ -130,27 +138,22 @@ static	void	 buf_alloc(char **, size_t *, size_t);
 static	void	 buf_dup(struct mchars *, char **, const char *);
 static	void	 buf_redup(struct mchars *, char **, 
 			size_t *, const char *);
-static	void	 error(const char *, ...);
 static	int	 sort_cat(const void *, const void *);
 static	int	 sort_title(const void *, const void *);
-static	void	 state_destroy(struct state *);
-static	int	 state_getrecord(struct state *, recno_t, struct rec *);
-static	int	 state_init(struct state *, 
-			const char *, const char *,
-			void (*err)(const char *),
-			void (*errx)(const char *, ...));
+static	int	 state_getrecord(struct state *, 
+			recno_t, struct rec *);
 static	void	 state_output(const struct res *, int);
-static	void	 state_search(struct state *, 
+static	int	 state_search(struct state *, 
 			const struct opts *, char *);
-
 static	void	 usage(void);
 
-static	const char	 *progname;
+static	char	*progname;
 
 int
 apropos(int argc, char *argv[])
 {
-	int		 ch, i;
+	BTREEINFO	 info;
+	int		 ch, i, rc;
 	const char	*dbf, *idxf;
 	struct state	 state;
 	char		*q, *v;
@@ -159,10 +162,12 @@ apropos(int argc, char *argv[])
 	extern char	*optarg;
 
 	memset(&opts, 0, sizeof(struct opts));
+	memset(&state, 0, sizeof(struct state));
 
 	dbf = "mandoc.db";
 	idxf = "mandoc.index";
 	q = NULL;
+	rc = EXIT_FAILURE;
 
 	progname = strrchr(argv[0], '/');
 	if (progname == NULL)
@@ -200,7 +205,7 @@ apropos(int argc, char *argv[])
 			if (i < SORT__MAX)
 				break;
 
-			error("%s: Bad sort\n", optarg);
+			fprintf(stderr, "%s: Bad sort\n", optarg);
 			return(EXIT_FAILURE);
 		case ('t'):
 			while (NULL != (v = strsep(&optarg, ","))) {
@@ -218,7 +223,7 @@ apropos(int argc, char *argv[])
 			if (NULL == v)
 				break;
 			
-			error("%s: Bad type\n", v);
+			fprintf(stderr, "%s: Bad type\n", v);
 			return(EXIT_FAILURE);
 		default:
 			usage();
@@ -230,54 +235,85 @@ apropos(int argc, char *argv[])
 
 	if (0 == argc || '\0' == **argv) {
 		usage();
-		return(EXIT_FAILURE);
+		goto out;
 	} else
 		q = *argv;
 
 	if (0 == opts.types)
 		opts.types = TYPE_NAME | TYPE_DESC;
 
-	if ( ! state_init(&state, dbf, idxf, perror, error)) {
-		state_destroy(&state);
-		return(EXIT_FAILURE);
+	/*
+	 * Configure databases.
+	 * The keyword database is a btree that allows for duplicate
+	 * entries.
+	 * The index database is a recno.
+	 */
+
+	memset(&info, 0, sizeof(BTREEINFO));
+	info.flags = R_DUP;
+
+	state.db = dbopen(dbf, O_RDONLY, 0, DB_BTREE, &info);
+	if (NULL == state.db) {
+		perror(dbf);
+		goto out;
+	}
+
+	state.idx = dbopen(idxf, O_RDONLY, 0, DB_RECNO, NULL);
+	if (NULL == state.idx) {
+		perror(idxf);
+		goto out;
 	}
 
-	state_search(&state, &opts, q);
-	state_destroy(&state);
+	/* Main search function. */
 
-	return(EXIT_SUCCESS);
+	rc = state_search(&state, &opts, q) ?
+		EXIT_SUCCESS : EXIT_FAILURE;
+out:
+	if (state.db)
+		(*state.db->close)(state.db);
+	if (state.idx)
+		(*state.idx->close)(state.idx);
+
+	return(rc);
 }
 
-static void
+static int
 state_search(struct state *p, const struct opts *opts, char *q)
 {
-	int		 i, len, ch, rflags, dflag;
+	int		 leaf, root, len, ch, dflag, rc;
 	struct mchars	*mc;
 	char		*buf;
 	size_t		 bufsz;
 	recno_t		 rec;
 	uint32_t	 fl;
 	DBT		 key, val;
-	struct res	 res[MAXRESULTS];
+	struct res	*res;
 	regex_t		 reg;
 	regex_t		*regp;
 	char		 filebuf[10];
 	struct rec	 record;
 
+	rc = 0;
+	root = leaf = -1;
+	res = NULL;
 	len = 0;
 	buf = NULL;
 	bufsz = 0;
-	ch = 0;
 	regp = NULL;
 
+	/*
+	 * Configure how we scan through results to see if we match:
+	 * whether by regexp or exact matches.
+	 */
+
 	switch (opts->match) {
 	case (MATCH_REGEX):
-		rflags = REG_EXTENDED | REG_NOSUB | 
+		ch = REG_EXTENDED | REG_NOSUB | 
 			(opts->insens ? REG_ICASE : 0);
 
-		if (0 != regcomp(&reg, q, rflags)) {
-			error("%s: Bad pattern\n", q);
-			return;
+		if (0 != regcomp(&reg, q, ch)) {
+			fprintf(stderr, "%s: Bad pattern\n", q);
+			return(0);
 		}
 
 		regp = &reg;
@@ -293,10 +329,7 @@ state_search(struct state *p, const struct opts *opts, char *q)
 		break;
 	}
 
-	if (NULL == (mc = mchars_alloc())) {
-		perror(NULL);
-		exit(EXIT_FAILURE);
-	}
+	mc = mchars_alloc();
 
 	/*
 	 * Iterate over the entire keyword database.
@@ -305,10 +338,7 @@ state_search(struct state *p, const struct opts *opts, char *q)
 	 * Lastly, add it to the available records.
 	 */
 
-	while (len < MAXRESULTS) {
-		if ((ch = (*p->db->seq)(p->db, &key, &val, dflag)))
-			break;
-
+	while (0 == (ch = (*p->db->seq)(p->db, &key, &val, dflag))) {
 		dflag = R_NEXT;
 
 		/* 
@@ -319,8 +349,8 @@ state_search(struct state *p, const struct opts *opts, char *q)
 		 */
 
 		if (key.size < 2 || 8 != val.size) {
-			error("%s: Corrupt database\n", p->dbf);
-			exit(EXIT_FAILURE);
+			fprintf(stderr, "%s: Bad database\n", p->dbf);
+			goto out;
 		}
 
 		buf_redup(mc, &buf, &bufsz, (char *)key.data);
@@ -357,7 +387,7 @@ state_search(struct state *p, const struct opts *opts, char *q)
 		memcpy(&rec, val.data + 4, sizeof(recno_t));
 
 		if ( ! state_getrecord(p, rec, &record))
-			exit(EXIT_FAILURE);
+			goto out;
 
 		/* If we're in a different section, skip... */
 
@@ -366,15 +396,25 @@ state_search(struct state *p, const struct opts *opts, char *q)
 		if (opts->arch && strcasecmp(opts->arch, record.arch))
 			continue;
 
-		/* FIXME: this needs to be changed.  Ugh.  Linear. */
+		/* 
+		 * Do a binary search to dedupe the results tree of the
+		 * same record: we don't print the same file.
+		 */
 
-		for (i = 0; i < len; i++)
-			if (res[i].rec == record.rec)
+		for (leaf = root; leaf >= 0; )
+			if (rec > res[leaf].rec && res[leaf].rhs >= 0)
+				leaf = res[leaf].rhs;
+			else if (rec < res[leaf].rec && res[leaf].lhs >= 0)
+				leaf = res[leaf].lhs;
+			else 
 				break;
 
-		if (i < len)
+		if (leaf >= 0 && res[leaf].rec == rec)
 			continue;
 
+		res = mandoc_realloc
+			(res, (len + 1) * sizeof(struct res));
+
 		/*
 		 * Now we have our filename, keywords, types, and all
 		 * other necessary information.  
@@ -387,6 +427,7 @@ state_search(struct state *p, const struct opts *opts, char *q)
 
 		res[len].rec = record.rec;
 		res[len].types = fl;
+		res[len].lhs = res[len].rhs = -1;
 
 		buf_dup(mc, &res[len].keyword, buf);
 		buf_dup(mc, &res[len].uri, filebuf);
@@ -394,26 +435,33 @@ state_search(struct state *p, const struct opts *opts, char *q)
 		buf_dup(mc, &res[len].arch, record.arch);
 		buf_dup(mc, &res[len].title, record.title);
 		buf_dup(mc, &res[len].desc, record.desc);
+
+		if (leaf >= 0) {
+			if (record.rec > res[leaf].rec)
+				res[leaf].rhs = len;
+			else
+				res[leaf].lhs = len;
+		} else
+			root = len;
+
 		len++;
 	}
 
-send:
 	if (ch < 0) {
 		perror(p->dbf);
-		exit(EXIT_FAILURE);
+		goto out;
 	} 
+send:
+	/* Sort our results. */
 
-	switch (opts->sort) {
-	case (SORT_CAT):
+	if (SORT_CAT == opts->sort)
 		qsort(res, len, sizeof(struct res), sort_cat);
-		break;
-	default:
+	else
 		qsort(res, len, sizeof(struct res), sort_title);
-		break;
-	}
 
 	state_output(res, len);
-
+	rc = 1;
+out:
 	for (len-- ; len >= 0; len--) {
 		free(res[len].keyword);
 		free(res[len].title);
@@ -423,11 +471,14 @@ send:
 		free(res[len].uri);
 	}
 
+	free(res);
 	free(buf);
 	mchars_free(mc);
 
 	if (regp)
 		regfree(regp);
+
+	return(rc);
 }
 
 /*
@@ -441,10 +492,7 @@ buf_alloc(char **buf, size_t *bufsz, size_t sz)
 		return;
 
 	*bufsz = sz + 1024;
-	if (NULL == (*buf = realloc(*buf, *bufsz))) {
-		perror(NULL);
-		exit(EXIT_FAILURE);
-	}
+	*buf = mandoc_realloc(*buf, *bufsz);
 }
 
 /*
@@ -535,16 +583,6 @@ buf_redup(struct mchars *mc, char **buf,
 }
 
 static void
-error(const char *fmt, ...)
-{
-	va_list		 ap;
-
-	va_start(ap, fmt);
-	vfprintf(stderr, fmt, ap);
-	va_end(ap);
-}
-
-static void
 state_output(const struct res *res, int sz)
 {
 	int		 i;
@@ -571,48 +609,6 @@ usage(void)
 }
 
 static int
-state_init(struct state *p, 
-		const char *dbf, const char *idxf,
-		void (*err)(const char *),
-		void (*errx)(const char *, ...))
-{
-	BTREEINFO	 info;
-
-	memset(p, 0, sizeof(struct state));
-	memset(&info, 0, sizeof(BTREEINFO));
-
-	info.flags = R_DUP;
-
-	p->dbf = dbf;
-	p->idxf = idxf;
-	p->err = err;
-
-	p->db = dbopen(p->dbf, O_RDONLY, 0, DB_BTREE, &info);
-	if (NULL == p->db) {
-		(*err)(p->dbf);
-		return(0);
-	}
-
-	p->idx = dbopen(p->idxf, O_RDONLY, 0, DB_RECNO, NULL);
-	if (NULL == p->idx) {
-		(*err)(p->idxf);
-		return(0);
-	}
-
-	return(1);
-}
-
-static void
-state_destroy(struct state *p)
-{
-
-	if (p->db)
-		(*p->db->close)(p->db);
-	if (p->idx)
-		(*p->idx->close)(p->idx);
-}
-
-static int
 state_getrecord(struct state *p, recno_t rec, struct rec *rp)
 {
 	DBT		key, val;
@@ -624,40 +620,33 @@ state_getrecord(struct state *p, recno_t rec, struct rec *rp)
 
 	rc = (*p->idx->get)(p->idx, &key, &val, 0);
 	if (rc < 0) {
-		(*p->err)(p->idxf);
-		return(0);
-	} else if (rc > 0) {
-		(*p->errx)("%s: Corrupt index\n", p->idxf);
+		perror(p->idxf);
 		return(0);
-	}
+	} else if (rc > 0)
+		goto err;
 
 	rp->file = (char *)val.data;
-	if ((sz = strlen(rp->file) + 1) >= val.size) {
-		(*p->errx)("%s: Corrupt index\n", p->idxf);
-		return(0);
-	}
+	if ((sz = strlen(rp->file) + 1) >= val.size)
+		goto err;
 
 	rp->cat = (char *)val.data + (int)sz;
-	if ((sz += strlen(rp->cat) + 1) >= val.size) {
-		(*p->errx)("%s: Corrupt index\n", p->idxf);
-		return(0);
-	}
+	if ((sz += strlen(rp->cat) + 1) >= val.size)
+		goto err;
 
 	rp->title = (char *)val.data + (int)sz;
-	if ((sz += strlen(rp->title) + 1) >= val.size) {
-		(*p->errx)("%s: Corrupt index\n", p->idxf);
-		return(0);
-	}
+	if ((sz += strlen(rp->title) + 1) >= val.size)
+		goto err;
 
 	rp->arch = (char *)val.data + (int)sz;
-	if ((sz += strlen(rp->arch) + 1) >= val.size) {
-		(*p->errx)("%s: Corrupt index\n", p->idxf);
-		return(0);
-	}
+	if ((sz += strlen(rp->arch) + 1) >= val.size)
+		goto err;
 
 	rp->desc = (char *)val.data + (int)sz;
 	rp->rec = rec;
 	return(1);
+err:
+	fprintf(stderr, "%s: Corrupt index\n", p->idxf);
+	return(0);
 }
 
 static int
diff --git a/usr.bin/mandoc/chars.in b/usr.bin/mandoc/chars.in
index 357f7dba95a..e3bdbbfd760 100644
--- a/usr.bin/mandoc/chars.in
+++ b/usr.bin/mandoc/chars.in
@@ -1,4 +1,4 @@
-/*	$Id: chars.in,v 1.17 2011/09/18 10:25:28 schwarze Exp $ */
+/*	$Id: chars.in,v 1.18 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -33,15 +33,15 @@ static const char ascii_nbrsp[2] = { ASCII_NBRSP, '\0' };
 CHAR_TBL_START
 
 /* Spacing. */
-CHAR("c",			"",		8203)
+CHAR("c",			"",		0)
 CHAR("0",			" ",		8194)
 CHAR(" ",			ascii_nbrsp,	160)
 CHAR("~",			ascii_nbrsp,	160)
-CHAR("%",			"",		8203)
-CHAR("&",			"",		8203)
-CHAR("^",			"",		8203)
-CHAR("|",			"",		8203)
-CHAR("}",			"",		8203)
+CHAR("%",			"",		0)
+CHAR("&",			"",		0)
+CHAR("^",			"",		0)
+CHAR("|",			"",		0)
+CHAR("}",			"",		0)
 
 /* Accents. */
 CHAR("a\"",			"\"",		779)
diff --git a/usr.bin/mandoc/html.c b/usr.bin/mandoc/html.c
index 0f719dfe823..80f6acc6956 100644
--- a/usr.bin/mandoc/html.c
+++ b/usr.bin/mandoc/html.c
@@ -1,4 +1,4 @@
-/*	$Id: html.c,v 1.28 2011/07/08 17:47:54 schwarze Exp $ */
+/*	$Id: html.c,v 1.29 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2011 Ingo Schwarze <schwarze@openbsd.org>
@@ -114,13 +114,14 @@ static void *
 ml_alloc(char *outopts, enum htmltype type)
 {
 	struct html	*h;
-	const char	*toks[4];
+	const char	*toks[5];
 	char		*v;
 
 	toks[0] = "style";
 	toks[1] = "man";
 	toks[2] = "includes";
-	toks[3] = NULL;
+	toks[3] = "fragment";
+	toks[4] = NULL;
 
 	h = mandoc_calloc(1, sizeof(struct html));
 
@@ -139,6 +140,9 @@ ml_alloc(char *outopts, enum htmltype type)
 		case (2):
 			h->base_includes = v;
 			break;
+		case (3):
+			h->oflags |= HTML_FRAGMENT;
+			break;
 		default:
 			break;
 		}
diff --git a/usr.bin/mandoc/html.h b/usr.bin/mandoc/html.h
index 8e86973f2b0..bcc6d6f7c39 100644
--- a/usr.bin/mandoc/html.h
+++ b/usr.bin/mandoc/html.h
@@ -1,4 +1,4 @@
-/*	$Id: html.h,v 1.17 2011/09/18 10:25:28 schwarze Exp $ */
+/*	$Id: html.h,v 1.18 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -104,7 +104,7 @@ struct	htmlpair {
 #define	PAIR_STYLE_INIT(p, h)	PAIR_INIT(p, ATTR_STYLE, (h)->buf)
 #define	PAIR_SUMMARY_INIT(p, v)	PAIR_INIT(p, ATTR_SUMMARY, v)
 
-enum	htmltype {
+enum	htmltype { 
 	HTML_HTML_4_01_STRICT,
 	HTML_XHTML_1_0_STRICT
 };
@@ -125,11 +125,13 @@ struct	html {
 	char		 *base_includes; /* base for include href */
 	char		 *style; /* style-sheet URI */
 	char		  buf[BUFSIZ]; /* see bufcat and friends */
-	size_t		  buflen;
+	size_t		  buflen; 
 	struct tag	 *metaf; /* current open font scope */
 	enum htmlfont	  metal; /* last used font */
 	enum htmlfont	  metac; /* current font mode */
-	enum htmltype	  type;
+	enum htmltype	  type; /* output media type */
+	int		  oflags; /* output options */
+#define	HTML_FRAGMENT	 (1 << 0) /* don't emit HTML/HEAD/BODY */
 };
 
 void		  print_gen_decls(struct html *);
diff --git a/usr.bin/mandoc/main.c b/usr.bin/mandoc/main.c
index 5c858a809c5..f8d4de1e58c 100644
--- a/usr.bin/mandoc/main.c
+++ b/usr.bin/mandoc/main.c
@@ -1,4 +1,4 @@
-/*	$Id: main.c,v 1.79 2011/10/06 23:04:16 schwarze Exp $ */
+/*	$Id: main.c,v 1.80 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>
@@ -126,6 +126,12 @@ main(int argc, char *argv[])
 
 	curp.mp = mparse_alloc(type, curp.wlevel, mmsg, &curp);
 
+	/*
+	 * Conditionally start up the lookaside buffer before parsing.
+	 */
+	if (OUTT_MAN == curp.outtype)
+		mparse_keep(curp.mp);
+
 	argc -= optind;
 	argv += optind;
 
@@ -251,6 +257,7 @@ parse(struct curparse *curp, int fd,
 			break;
 		case (OUTT_MAN):
 			curp->outmdoc = man_mdoc;
+			curp->outman = man_man;
 			break;
 		case (OUTT_PDF):
 			/* FALLTHROUGH */
diff --git a/usr.bin/mandoc/main.h b/usr.bin/mandoc/main.h
index 88916340d52..f6be7534525 100644
--- a/usr.bin/mandoc/main.h
+++ b/usr.bin/mandoc/main.h
@@ -1,4 +1,4 @@
-/*	$Id: main.h,v 1.9 2011/09/17 14:45:22 schwarze Exp $ */
+/*	$Id: main.h,v 1.10 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -42,6 +42,7 @@ void		  tree_mdoc(void *, const struct mdoc *);
 void		  tree_man(void *, const struct man *);
 
 void		  man_mdoc(void *, const struct mdoc *);
+void		  man_man(void *, const struct man *);
 
 void		 *locale_alloc(char *);
 void		 *utf8_alloc(char *);
diff --git a/usr.bin/mandoc/man.c b/usr.bin/mandoc/man.c
index 35eaf6ab72b..3ce1b5d0b79 100644
--- a/usr.bin/mandoc/man.c
+++ b/usr.bin/mandoc/man.c
@@ -1,4 +1,4 @@
-/*	$Id: man.c,v 1.61 2011/09/18 15:54:48 schwarze Exp $ */
+/*	$Id: man.c,v 1.62 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -644,3 +644,11 @@ man_node_unlink(struct man *m, struct man_node *n)
 	if (m && m->first == n)
 		m->first = NULL;
 }
+
+const struct mparse *
+man_mparse(const struct man *m)
+{
+
+	assert(m && m->parse);
+	return(m->parse);
+}
diff --git a/usr.bin/mandoc/man.h b/usr.bin/mandoc/man.h
index 4595ec66fec..4a077935406 100644
--- a/usr.bin/mandoc/man.h
+++ b/usr.bin/mandoc/man.h
@@ -1,4 +1,4 @@
-/*	$Id: man.h,v 1.37 2011/04/24 16:22:02 schwarze Exp $ */
+/*	$Id: man.h,v 1.38 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -105,6 +105,7 @@ struct	man;
 
 const struct man_node *man_node(const struct man *);
 const struct man_meta *man_meta(const struct man *);
+const struct mparse   *man_mparse(const struct man *);
 
 __END_DECLS
 
diff --git a/usr.bin/mandoc/man_html.c b/usr.bin/mandoc/man_html.c
index 5192ba9761e..0e44a8074bf 100644
--- a/usr.bin/mandoc/man_html.c
+++ b/usr.bin/mandoc/man_html.c
@@ -1,4 +1,4 @@
-/*	$Id: man_html.c,v 1.42 2011/09/18 15:54:48 schwarze Exp $ */
+/*	$Id: man_html.c,v 1.43 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -137,33 +137,32 @@ print_bvspace(struct html *h, const struct man_node *n)
 void
 html_man(void *arg, const struct man *m)
 {
-	struct html	*h;
-	struct tag	*t;
 	struct mhtml	 mh;
 
-	h = (struct html *)arg;
-
-	print_gen_decls(h);
-
 	memset(&mh, 0, sizeof(struct mhtml));
-
-	t = print_otag(h, TAG_HTML, 0, NULL);
-	print_man(man_meta(m), man_node(m), &mh, h);
-	print_tagq(h, t);
-
-	printf("\n");
+	print_man(man_meta(m), man_node(m), &mh, (struct html *)arg);
+	putchar('\n');
 }
 
 static void
 print_man(MAN_ARGS) 
 {
-	struct tag	*t;
+	struct tag	*t, *tt;
+	struct htmlpair	 tag;
 
-	t = print_otag(h, TAG_HEAD, 0, NULL);
-	print_man_head(m, n, mh, h);
-	print_tagq(h, t);
+	PAIR_CLASS_INIT(&tag, "mandoc");
+
+	if ( ! (HTML_FRAGMENT & h->oflags)) {
+		print_gen_decls(h);
+		t = print_otag(h, TAG_HTML, 0, NULL);
+		tt = print_otag(h, TAG_HEAD, 0, NULL);
+		print_man_head(m, n, mh, h);
+		print_tagq(h, tt);
+		print_otag(h, TAG_BODY, 0, NULL);
+		print_otag(h, TAG_DIV, 1, &tag);
+	} else 
+		t = print_otag(h, TAG_DIV, 1, &tag);
 
-	t = print_otag(h, TAG_BODY, 0, NULL);
 	print_man_nodelist(m, n, mh, h);
 	print_tagq(h, t);
 }
diff --git a/usr.bin/mandoc/mandoc.1 b/usr.bin/mandoc/mandoc.1
index 842f6f72a27..914c2a9b7e8 100644
--- a/usr.bin/mandoc/mandoc.1
+++ b/usr.bin/mandoc/mandoc.1
@@ -1,4 +1,4 @@
-.\"	$OpenBSD: mandoc.1,v 1.48 2011/09/26 14:49:03 jmc Exp $
+.\"	$OpenBSD: mandoc.1,v 1.49 2011/10/09 17:59:56 schwarze Exp $
 .\"
 .\" Copyright (c) 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
 .\"
@@ -14,7 +14,7 @@
 .\" ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 .\" OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 .\"
-.Dd $Mdocdate: September 26 2011 $
+.Dd $Mdocdate: October 9 2011 $
 .Dt MANDOC 1
 .Os
 .Sh NAME
@@ -176,11 +176,11 @@ Encode output using the current locale.
 See
 .Sx Locale Output .
 .It Fl T Ns Cm man
-Produce output in
+Produce
 .Xr man 7
-format; only useful when applied to
-.Fl m Ns Cm doc
-input.
+format output.
+See
+.Sx Man Output .
 .It Fl T Ns Cm pdf
 Produce PDF output.
 See
@@ -259,6 +259,12 @@ The following
 .Fl O
 arguments are accepted:
 .Bl -tag -width Ds
+.It Cm fragment
+Do not emit doctype, html, and body elements.
+The
+.Cm style
+argument shall be unused.
+This is useful when embedding manual content within existing documents.
 .It Cm includes Ns = Ns Ar fmt
 The string
 .Ar fmt ,
@@ -305,6 +311,26 @@ will fall back to
 See
 .Sx ASCII Output
 for font style specification and available command-line arguments.
+.Ss Man Output
+Translate input format into
+.Xr man 7
+output format.
+This is useful for distributing manual sources to legancy systems
+lacking
+.Xr mdoc 7
+formatters.
+.Pp
+If
+.Xr mdoc 7
+is passed as input, it is translated into
+.Xr man 7 ;
+if the input format is
+.Xr man 7 ,
+it is parsed and re-outputted.
+In either case, the
+.Xr roff 7
+.Sq so
+macros are processed prior to producing output.
 .Ss PDF Output
 PDF-1.1 output may be generated by
 .Fl T Ns Cm pdf .
diff --git a/usr.bin/mandoc/mandoc.h b/usr.bin/mandoc/mandoc.h
index 473d80253f1..cef90417ad9 100644
--- a/usr.bin/mandoc/mandoc.h
+++ b/usr.bin/mandoc/mandoc.h
@@ -1,4 +1,4 @@
-/*	$Id: mandoc.h,v 1.40 2011/09/18 10:25:28 schwarze Exp $ */
+/*	$Id: mandoc.h,v 1.41 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -400,30 +400,31 @@ struct	man;
 
 __BEGIN_DECLS
 
-void		  mparse_free(struct mparse *);
-void		  mparse_reset(struct mparse *);
-struct mparse	 *mparse_alloc(enum mparset, 
-			enum mandoclevel, mandocmsg, void *);
-enum mandoclevel  mparse_readfd(struct mparse *, int, const char *);
-void		  mparse_result(struct mparse *, struct mdoc **, struct man **);
-const char	 *mparse_strerror(enum mandocerr);
-const char	 *mparse_strlevel(enum mandoclevel);
-
 void		 *mandoc_calloc(size_t, size_t);
+enum mandoc_esc	  mandoc_escape(const char **, const char **, int *);
 void		 *mandoc_malloc(size_t);
 void		 *mandoc_realloc(void *, size_t);
 char		 *mandoc_strdup(const char *);
 char		 *mandoc_strndup(const char *, size_t);
-
-enum mandoc_esc	  mandoc_escape(const char **, const char **, int *);
-
 struct mchars	 *mchars_alloc(void);
+void		  mchars_free(struct mchars *);
 char	 	  mchars_num2char(const char *, size_t);
 int		  mchars_num2uc(const char *, size_t);
-const char	 *mchars_spec2str(struct mchars *, const char *, size_t, size_t *);
-int		  mchars_spec2cp(struct mchars *, const char *, size_t);
-void		  mchars_free(struct mchars *);
-
+int		  mchars_spec2cp(struct mchars *, 
+			const char *, size_t);
+const char	 *mchars_spec2str(struct mchars *, 
+			const char *, size_t, size_t *);
+struct mparse	 *mparse_alloc(enum mparset, 
+			enum mandoclevel, mandocmsg, void *);
+void		  mparse_free(struct mparse *);
+void		  mparse_keep(struct mparse *);
+enum mandoclevel  mparse_readfd(struct mparse *, int, const char *);
+void		  mparse_reset(struct mparse *);
+void		  mparse_result(struct mparse *, 
+			struct mdoc **, struct man **);
+const char	 *mparse_getkeep(const struct mparse *);
+const char	 *mparse_strerror(enum mandocerr);
+const char	 *mparse_strlevel(enum mandoclevel);
 
 __END_DECLS
 
diff --git a/usr.bin/mandoc/mandocdb.8 b/usr.bin/mandoc/mandocdb.8
index 12602b648fa..1602aa06cfa 100644
--- a/usr.bin/mandoc/mandocdb.8
+++ b/usr.bin/mandoc/mandocdb.8
@@ -1,4 +1,4 @@
-.\"	$Id: mandocdb.8,v 1.3 2011/09/18 10:25:28 schwarze Exp $
+.\"	$Id: mandocdb.8,v 1.4 2011/10/09 17:59:56 schwarze Exp $
 .\"
 .\" Copyright (c) 2011 Kristaps Dzonsons <kristaps@bsd.lv>
 .\"
@@ -14,7 +14,7 @@
 .\" ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 .\" OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 .\"
-.Dd $Mdocdate: September 18 2011 $
+.Dd $Mdocdate: October 9 2011 $
 .Dt MANDOCDB 8
 .Os
 .Sh NAME
@@ -45,12 +45,12 @@ for fast retrieval.
 The arguments are as follows:
 .Bl -tag -width Ds
 .It Fl d Ar dir
-Remove
+Merge (remove and re-add)
 .Ar
 from the databases in
 .Ar dir .
 .It Fl u Ar dir
-Update (remove and re-add)
+Remove
 .Ar
 from the databases in
 .Ar dir .
diff --git a/usr.bin/mandoc/mdoc_html.c b/usr.bin/mandoc/mdoc_html.c
index c3ab375ce1b..680c1cc53c7 100644
--- a/usr.bin/mandoc/mdoc_html.c
+++ b/usr.bin/mandoc/mdoc_html.c
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_html.c,v 1.61 2011/09/18 15:54:48 schwarze Exp $ */
+/*	$Id: mdoc_html.c,v 1.62 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  *
@@ -258,17 +258,9 @@ static	const char * const lists[LIST_MAX] = {
 void
 html_mdoc(void *arg, const struct mdoc *m)
 {
-	struct html 	*h;
-	struct tag	*t;
-
-	h = (struct html *)arg;
-
-	print_gen_decls(h);
-	t = print_otag(h, TAG_HTML, 0, NULL);
-	print_mdoc(mdoc_meta(m), mdoc_node(m), h);
-	print_tagq(h, t);
 
-	printf("\n");
+	print_mdoc(mdoc_meta(m), mdoc_node(m), (struct html *)arg);
+	putchar('\n');
 }
 
 
@@ -356,13 +348,22 @@ a2offs(const char *p, struct roffsu *su)
 static void
 print_mdoc(MDOC_ARGS)
 {
-	struct tag	*t;
+	struct tag	*t, *tt;
+	struct htmlpair	 tag;
 
-	t = print_otag(h, TAG_HEAD, 0, NULL);
-	print_mdoc_head(m, n, h);
-	print_tagq(h, t);
+	PAIR_CLASS_INIT(&tag, "mandoc");
+
+	if ( ! (HTML_FRAGMENT & h->oflags)) {
+		print_gen_decls(h);
+		t = print_otag(h, TAG_HTML, 0, NULL);
+		tt = print_otag(h, TAG_HEAD, 0, NULL);
+		print_mdoc_head(m, n, h);
+		print_tagq(h, tt);
+		print_otag(h, TAG_BODY, 0, NULL);
+		print_otag(h, TAG_DIV, 1, &tag);
+	} else 
+		t = print_otag(h, TAG_DIV, 1, &tag);
 
-	t = print_otag(h, TAG_BODY, 0, NULL);
 	print_mdoc_nodelist(m, n, h);
 	print_tagq(h, t);
 }
diff --git a/usr.bin/mandoc/mdoc_man.c b/usr.bin/mandoc/mdoc_man.c
index 3844eb36dea..7ba8ed6e539 100644
--- a/usr.bin/mandoc/mdoc_man.c
+++ b/usr.bin/mandoc/mdoc_man.c
@@ -1,4 +1,4 @@
-/*	$Id: mdoc_man.c,v 1.3 2011/09/30 00:13:21 schwarze Exp $ */
+/*	$Id: mdoc_man.c,v 1.4 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2011 Ingo Schwarze <schwarze@openbsd.org>
  *
@@ -18,50 +18,51 @@
 #include <string.h>
 
 #include "mandoc.h"
+#include "man.h"
 #include "mdoc.h"
 #include "main.h"
 
-static	int	  need_space = 0;
-static	int	  need_nl = 0;
-
 #define	DECL_ARGS const struct mdoc_meta *m, \
-		  const struct mdoc_node *n
+		  const struct mdoc_node *n, \
+		  struct mman *mm
 
-struct	manact {
-	int		(*cond)(DECL_ARGS);
-	int		(*pre)(DECL_ARGS);
-	void		(*post)(DECL_ARGS);
-	const char	 *prefix;
-	const char	 *suffix;
+struct	mman {
+	int		  need_space; /* next word needs prior ws */
+	int		  need_nl; /* next word needs prior nl */
 };
 
-static	void	  print_word(const char *);
-static	void	  print_node(DECL_ARGS);
+struct	manact {
+	int		(*cond)(DECL_ARGS); /* DON'T run actions */
+	int		(*pre)(DECL_ARGS); /* pre-node action */
+	void		(*post)(DECL_ARGS); /* post-node action */
+	const char	 *prefix; /* pre-node string constant */
+	const char	 *suffix; /* post-node string constant */
+};
 
-static	int	  cond_head(DECL_ARGS);
 static	int	  cond_body(DECL_ARGS);
-static	int	  pre_enc(DECL_ARGS);
+static	int	  cond_head(DECL_ARGS);
+static	void	  post_bd(DECL_ARGS);
+static	void	  post_dl(DECL_ARGS);
 static	void	  post_enc(DECL_ARGS);
+static	void	  post_nm(DECL_ARGS);
 static	void	  post_percent(DECL_ARGS);
-static	int	  pre_sect(DECL_ARGS);
+static	void	  post_pf(DECL_ARGS);
 static	void	  post_sect(DECL_ARGS);
-
+static	void	  post_sp(DECL_ARGS);
 static	int	  pre_ap(DECL_ARGS);
 static	int	  pre_bd(DECL_ARGS);
-static	void	  post_bd(DECL_ARGS);
 static	int	  pre_br(DECL_ARGS);
 static	int	  pre_dl(DECL_ARGS);
-static	void	  post_dl(DECL_ARGS);
+static	int	  pre_enc(DECL_ARGS);
 static	int	  pre_it(DECL_ARGS);
 static	int	  pre_nm(DECL_ARGS);
-static	void	  post_nm(DECL_ARGS);
 static	int	  pre_ns(DECL_ARGS);
-static	void	  post_pf(DECL_ARGS);
 static	int	  pre_pp(DECL_ARGS);
 static	int	  pre_sp(DECL_ARGS);
-static	void	  post_sp(DECL_ARGS);
+static	int	  pre_sect(DECL_ARGS);
 static	int	  pre_xr(DECL_ARGS);
-
+static	void	  print_word(struct mman *, const char *);
+static	void	  print_node(DECL_ARGS);
 
 static	const struct manact manacts[MDOC_MAX + 1] = {
 	{ NULL, pre_ap, NULL, NULL, NULL }, /* Ap */
@@ -102,7 +103,11 @@ static	const struct manact manacts[MDOC_MAX + 1] = {
 	{ cond_body, pre_enc, post_enc, "[", "]" }, /* Op */
 	{ NULL, NULL, NULL, NULL, NULL }, /* _Ot */
 	{ NULL, pre_enc, post_enc, "\\fI", "\\fP" }, /* _Pa */
-	{ NULL, NULL, NULL, NULL, NULL }, /* _Rv */
+	{ NULL, pre_enc, post_enc, "The \\fB",
+		"\\fP\nfunction returns the value 0 if successful;\n"
+		"otherwise the value -1 is returned and the global\n"
+		"variable \\fIerrno\\fP is set to indicate the error."
+		}, /* Rv */
 	{ NULL, NULL, NULL, NULL, NULL }, /* _St */
 	{ NULL, NULL, NULL, NULL, NULL }, /* _Va */
 	{ NULL, NULL, NULL, NULL, NULL }, /* _Vt */
@@ -191,18 +196,34 @@ static	const struct manact manacts[MDOC_MAX + 1] = {
 	{ NULL, NULL, NULL, NULL, NULL }, /* ROOT */
 };
 
-
 static void
-print_word(const char *s)
+print_word(struct mman *mm, const char *s)
 {
-	if (need_nl) {
+
+	if (mm->need_nl) {
+		/* 
+		 * If we need a newline, print it now and start afresh.
+		 */
 		putchar('\n');
-		need_space = 0;
-		need_nl = 0;
-	} else if (need_space &&
-	    (NULL == strchr(".,:;)]?!", s[0]) || '\0' != s[1]))
-		putchar(' ');
-	need_space = ('(' != s[0] && '[' != s[0]) || '\0' != s[1];
+		mm->need_space = 0;
+		mm->need_nl = 0;
+	} else if (mm->need_space && '\0' != s[0])
+		/*
+		 * If we need a space, only print it before
+		 * (1) a nonzero length word;
+		 * (2) a word that is non-punctuation; and
+		 * (3) if punctuation, non-terminating puncutation.
+		 */
+		if (NULL == strchr(".,:;)]?!", s[0]) || '\0' != s[1])
+			putchar(' ');
+
+	/*
+	 * Reassign needing space if we're not following opening
+	 * punctuation.
+	 */
+	mm->need_space = 
+		('(' != s[0] && '[' != s[0]) || '\0' != s[1];
+
 	for ( ; *s; s++) {
 		switch (*s) {
 		case (ASCII_NBRSP):
@@ -212,27 +233,42 @@ print_word(const char *s)
 			putchar('-');
 			break;
 		default:
-			putchar(*s);
+			putchar((unsigned char)*s);
 			break;
 		}
 	}
 }
 
 void
+man_man(void *arg, const struct man *man)
+{
+
+	/*
+	 * Dump the keep buffer.
+	 * We're guaranteed by now that this exists (is non-NULL).
+	 * Flush stdout afterward, just in case.
+	 */
+	fputs(mparse_getkeep(man_mparse(man)), stdout);
+	fflush(stdout);
+}
+
+void
 man_mdoc(void *arg, const struct mdoc *mdoc)
 {
 	const struct mdoc_meta *m;
 	const struct mdoc_node *n;
+	struct mman	        mm;
 
 	m = mdoc_meta(mdoc);
 	n = mdoc_node(mdoc);
 
 	printf(".TH \"%s\" \"%s\" \"%s\" \"%s\" \"%s\"",
-	    m->title, m->msec, m->date, m->os, m->vol);
-	need_nl = 1;
-	need_space = 0;
+			m->title, m->msec, m->date, m->os, m->vol);
 
-	print_node(m, n);
+	memset(&mm, 0, sizeof(struct mman));
+
+	mm.need_nl = 1;
+	print_node(m, n, &mm);
 	putchar('\n');
 }
 
@@ -240,61 +276,93 @@ static void
 print_node(DECL_ARGS)
 {
 	const struct mdoc_node	*prev, *sub;
-	const struct manact	*act = NULL;
+	const struct manact	*act;
 	int			 cond, do_sub;
-
+	
+	/*
+	 * Break the line if we were parsed subsequent the current node.
+	 * This makes the page structure be more consistent.
+	 */
 	prev = n->prev ? n->prev : n->parent;
 	if (prev && prev->line < n->line)
-		need_nl = 1;
+		mm->need_nl = 1;
 
+	act = NULL;
 	cond = 0;
 	do_sub = 1;
+
 	if (MDOC_TEXT == n->type) {
-		if (need_nl && ('.' == *n->string || '\'' == *n->string)) {
-			print_word("\\&");
-			need_space = 0;
+		/*
+		 * Make sure that we don't happen to start with a
+		 * control character at the start of a line.
+		 */
+		if (mm->need_nl && ('.' == *n->string || 
+					'\'' == *n->string)) {
+			print_word(mm, "\\&");
+			mm->need_space = 0;
 		}
-		print_word(n->string);
+		print_word(mm, n->string);
 	} else {
+		/*
+		 * Conditionally run the pre-node action handler for a
+		 * node.
+		 */
 		act = manacts + n->tok;
-		cond = NULL == act->cond || (*act->cond)(m, n);
+		cond = NULL == act->cond || (*act->cond)(m, n, mm);
 		if (cond && act->pre)
-			do_sub = (*act->pre)(m, n);
+			do_sub = (*act->pre)(m, n, mm);
 	}
 
+	/* 
+	 * Conditionally run all child nodes.
+	 * Note that this iterates over children instead of using
+	 * recursion.  This prevents unnecessary depth in the stack.
+	 */
 	if (do_sub)
 		for (sub = n->child; sub; sub = sub->next)
-			print_node(m, sub);
+			print_node(m, sub, mm);
 
+	/*
+	 * Lastly, conditionally run the post-node handler.
+	 */
 	if (cond && act->post)
-		(*act->post)(m, n);
+		(*act->post)(m, n, mm);
 }
 
 static int
 cond_head(DECL_ARGS)
 {
+
 	return(MDOC_HEAD == n->type);
 }
 
 static int
 cond_body(DECL_ARGS)
 {
+
 	return(MDOC_BODY == n->type);
 }
 
+/*
+ * Output a font encoding before a node, e.g., \fR.
+ * This obviously has no trailing space.
+ */
 static int
 pre_enc(DECL_ARGS)
 {
-	const char *prefix;
+	const char	*prefix;
 
 	prefix = manacts[n->tok].prefix;
 	if (NULL == prefix)
 		return(1);
-	print_word(prefix);
-	need_space = 0;
+	print_word(mm, prefix);
+	mm->need_space = 0;
 	return(1);
 }
 
+/*
+ * Output a font encoding subsequent a node, e.g., \fP.
+ */
 static void
 post_enc(DECL_ARGS)
 {
@@ -303,54 +371,65 @@ post_enc(DECL_ARGS)
 	suffix = manacts[n->tok].suffix;
 	if (NULL == suffix)
 		return;
-	need_space = 0;
-	print_word(suffix);
+	mm->need_space = 0;
+	print_word(mm, suffix);
 }
 
+/*
+ * Used in listings (percent = %A, e.g.).
+ * FIXME: this is incomplete. 
+ * It doesn't print a nice ", and" for lists.
+ */
 static void
 post_percent(DECL_ARGS)
 {
 
-	post_enc(m, n);
+	post_enc(m, n, mm);
 	if (n->next)
-		print_word(",");
+		print_word(mm, ",");
 	else {
-		print_word(".");
-		need_nl = 1;
+		print_word(mm, ".");
+		mm->need_nl = 1;
 	}
 }
 
+/*
+ * Print before a section header.
+ */
 static int
 pre_sect(DECL_ARGS)
 {
 
 	if (MDOC_HEAD != n->type)
 		return(1);
-	need_nl = 1;
-	print_word(manacts[n->tok].prefix);
-	print_word("\"");
-	need_space = 0;
+	mm->need_nl = 1;
+	print_word(mm, manacts[n->tok].prefix);
+	print_word(mm, "\"");
+	mm->need_space = 0;
 	return(1);
 }
 
+/*
+ * Print subsequent a section header.
+ */
 static void
 post_sect(DECL_ARGS)
 {
 
 	if (MDOC_HEAD != n->type)
 		return;
-	need_space = 0;
-	print_word("\"");
-	need_nl = 1;
+	mm->need_space = 0;
+	print_word(mm, "\"");
+	mm->need_nl = 1;
 }
 
 static int
 pre_ap(DECL_ARGS)
 {
 
-	need_space = 0;
-	print_word("'");
-	need_space = 0;
+	mm->need_space = 0;
+	print_word(mm, "'");
+	mm->need_space = 0;
 	return(0);
 }
 
@@ -360,10 +439,10 @@ pre_bd(DECL_ARGS)
 
 	if (DISP_unfilled == n->norm->Bd.type ||
 	    DISP_literal  == n->norm->Bd.type) {
-		need_nl = 1;
-		print_word(".nf");
+		mm->need_nl = 1;
+		print_word(mm, ".nf");
 	}
-	need_nl = 1;
+	mm->need_nl = 1;
 	return(1);
 }
 
@@ -373,19 +452,19 @@ post_bd(DECL_ARGS)
 
 	if (DISP_unfilled == n->norm->Bd.type ||
 	    DISP_literal  == n->norm->Bd.type) {
-		need_nl = 1;
-		print_word(".fi");
+		mm->need_nl = 1;
+		print_word(mm, ".fi");
 	}
-	need_nl = 1;
+	mm->need_nl = 1;
 }
 
 static int
 pre_br(DECL_ARGS)
 {
 
-	need_nl = 1;
-	print_word(".br");
-	need_nl = 1;
+	mm->need_nl = 1;
+	print_word(mm, ".br");
+	mm->need_nl = 1;
 	return(0);
 }
 
@@ -393,9 +472,9 @@ static int
 pre_dl(DECL_ARGS)
 {
 
-	need_nl = 1;
-	print_word(".RS 6n");
-	need_nl = 1;
+	mm->need_nl = 1;
+	print_word(mm, ".RS 6n");
+	mm->need_nl = 1;
 	return(1);
 }
 
@@ -403,9 +482,9 @@ static void
 post_dl(DECL_ARGS)
 {
 
-	need_nl = 1;
-	print_word(".RE");
-	need_nl = 1;
+	mm->need_nl = 1;
+	print_word(mm, ".RE");
+	mm->need_nl = 1;
 }
 
 static int
@@ -414,21 +493,21 @@ pre_it(DECL_ARGS)
 	const struct mdoc_node *bln;
 
 	if (MDOC_HEAD == n->type) {
-		need_nl = 1;
-		print_word(".TP");
+		mm->need_nl = 1;
+		print_word(mm, ".TP");
 		bln = n->parent->parent->prev;
 		switch (bln->norm->Bl.type) {
 		case (LIST_bullet):
-			print_word("4n");
-			need_nl = 1;
-			print_word("\\fBo\\fP");
+			print_word(mm, "4n");
+			mm->need_nl = 1;
+			print_word(mm, "\\fBo\\fP");
 			break;
 		default:
 			if (bln->norm->Bl.width)
-				print_word(bln->norm->Bl.width);
+				print_word(mm, bln->norm->Bl.width);
 			break;
 		}
-		need_nl = 1;
+		mm->need_nl = 1;
 	}
 	return(1);
 }
@@ -439,10 +518,10 @@ pre_nm(DECL_ARGS)
 
 	if (MDOC_ELEM != n->type && MDOC_HEAD != n->type)
 		return(1);
-	print_word("\\fB");
-	need_space = 0;
+	print_word(mm, "\\fB");
+	mm->need_space = 0;
 	if (NULL == n->child)
-		print_word(m->name);
+		print_word(mm, m->name);
 	return(1);
 }
 
@@ -452,15 +531,15 @@ post_nm(DECL_ARGS)
 
 	if (MDOC_ELEM != n->type && MDOC_HEAD != n->type)
 		return;
-	need_space = 0;
-	print_word("\\fP");
+	mm->need_space = 0;
+	print_word(mm, "\\fP");
 }
 
 static int
 pre_ns(DECL_ARGS)
 {
 
-	need_space = 0;
+	mm->need_space = 0;
 	return(0);
 }
 
@@ -468,19 +547,19 @@ static void
 post_pf(DECL_ARGS)
 {
 
-	need_space = 0;
+	mm->need_space = 0;
 }
 
 static int
 pre_pp(DECL_ARGS)
 {
 
-	need_nl = 1;
+	mm->need_nl = 1;
 	if (MDOC_It == n->parent->tok)
-		print_word(".sp");
+		print_word(mm, ".sp");
 	else
-		print_word(".PP");
-	need_nl = 1;
+		print_word(mm, ".PP");
+	mm->need_nl = 1;
 	return(1);
 }
 
@@ -488,8 +567,8 @@ static int
 pre_sp(DECL_ARGS)
 {
 
-	need_nl = 1;
-	print_word(".sp");
+	mm->need_nl = 1;
+	print_word(mm, ".sp");
 	return(1);
 }
 
@@ -497,7 +576,7 @@ static void
 post_sp(DECL_ARGS)
 {
 
-	need_nl = 1;
+	mm->need_nl = 1;
 }
 
 static int
@@ -507,13 +586,13 @@ pre_xr(DECL_ARGS)
 	n = n->child;
 	if (NULL == n)
 		return(0);
-	print_node(m, n);
+	print_node(m, n, mm);
 	n = n->next;
 	if (NULL == n)
 		return(0);
-	need_space = 0;
-	print_word("(");
-	print_node(m, n);
-	print_word(")");
+	mm->need_space = 0;
+	print_word(mm, "(");
+	print_node(m, n, mm);
+	print_word(mm, ")");
 	return(0);
 }
diff --git a/usr.bin/mandoc/read.c b/usr.bin/mandoc/read.c
index 1f31b7b27bb..aa31338153d 100644
--- a/usr.bin/mandoc/read.c
+++ b/usr.bin/mandoc/read.c
@@ -1,4 +1,4 @@
-/*	$Id: read.c,v 1.3 2011/09/18 10:25:28 schwarze Exp $ */
+/*	$Id: read.c,v 1.4 2011/10/09 17:59:56 schwarze Exp $ */
 /*
  * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
  * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>
@@ -53,6 +53,7 @@ struct	mparse {
 	mandocmsg	  mmsg; /* warning/error message handler */
 	void		 *arg; /* argument to mmsg */
 	const char	 *file; 
+	struct buf	 *secondary;
 };
 
 static	void	  resize_buf(struct buf *, size_t);
@@ -401,6 +402,27 @@ mparse_buf_r(struct mparse *curp, struct buf blk, int start)
 
 		of = 0;
 
+		/*
+		 * Maintain a lookaside buffer of all parsed lines.  We
+		 * only do this if mparse_keep() has been invoked (the
+		 * buffer may be accessed with mparse_getkeep()).
+		 */
+
+		if (curp->secondary) {
+			curp->secondary->buf = 
+				mandoc_realloc
+				(curp->secondary->buf, 
+				 curp->secondary->sz + pos + 2);
+			memcpy(curp->secondary->buf + 
+					curp->secondary->sz, 
+					ln.buf, pos);
+			curp->secondary->sz += pos;
+			curp->secondary->buf
+				[curp->secondary->sz] = '\n';
+			curp->secondary->sz++;
+			curp->secondary->buf
+				[curp->secondary->sz] = '\0';
+		}
 rerun:
 		rr = roff_parseln
 			(curp->roff, curp->line, 
@@ -427,6 +449,13 @@ rerun:
 			assert(MANDOCLEVEL_FATAL <= curp->file_status);
 			break;
 		case (ROFF_SO):
+			/*
+			 * We remove `so' clauses from our lookaside
+			 * buffer because we're going to descend into
+			 * the file recursively.
+			 */
+			if (curp->secondary) 
+				curp->secondary->sz -= pos + 1;
 			mparse_readfd_r(curp, -1, ln.buf + of, 1);
 			if (MANDOCLEVEL_FATAL <= curp->file_status)
 				break;
@@ -690,6 +719,8 @@ mparse_reset(struct mparse *curp)
 		mdoc_reset(curp->mdoc);
 	if (curp->man)
 		man_reset(curp->man);
+	if (curp->secondary)
+		curp->secondary->sz = 0;
 
 	curp->file_status = MANDOCLEVEL_OK;
 	curp->mdoc = NULL;
@@ -706,7 +737,10 @@ mparse_free(struct mparse *curp)
 		man_free(curp->pman);
 	if (curp->roff)
 		roff_free(curp->roff);
+	if (curp->secondary)
+		free(curp->secondary->buf);
 
+	free(curp->secondary);
 	free(curp);
 }
 
@@ -766,3 +800,19 @@ mparse_strlevel(enum mandoclevel lvl)
 {
 	return(mandoclevels[lvl]);
 }
+
+void
+mparse_keep(struct mparse *p)
+{
+
+	assert(NULL == p->secondary);
+	p->secondary = mandoc_calloc(1, sizeof(struct buf));
+}
+
+const char *
+mparse_getkeep(const struct mparse *p)
+{
+
+	assert(p->secondary);
+	return(p->secondary->sz ? p->secondary->buf : NULL);
+}