Using man_node_delete() instead of man_node_free()/man_node_freelist() and friends (much simpler).

Split blk_imp() into blk_exp() (explicit macros), blk_dotted() (roff macros), and the original.
Added de, dei, am, ami, and ig roff macros (for now, these are discarded within the parse).

8 files changed, 241 insertions, 51 deletions

diff --git a/libman.h b/libman.h
index d366d6f8..bee1b549 100644
--- a/libman.h
+++ b/libman.h
@@ -1,4 +1,4 @@
-/*	$Id: libman.h,v 1.26 2010/03/23 21:50:43 kristaps Exp $ */
+/*	$Id: libman.h,v 1.27 2010/03/24 20:10:53 kristaps Exp $ */
 /*
  * Copyright (c) 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -60,6 +60,7 @@ enum	merr {
 	WNOSCOPE,
 	WOLITERAL,
 	WNLITERAL,
+	WROFFNEST,
 	WERRMAX
 };
 
@@ -73,6 +74,7 @@ struct	man_macro {
 #define	MAN_EXPLICIT	 (1 << 1)	/* See blk_imp(). */
 #define	MAN_FSCOPED	 (1 << 2)	/* See blk_imp(). */
 #define	MAN_NSCOPED	 (1 << 3)	/* See in_line_eoln(). */
+#define	MAN_NOCLOSE	 (1 << 4)	/* See blk_exp(). */
 };
 
 extern	const struct man_macro *const man_macros;
@@ -93,9 +95,7 @@ int		  man_block_alloc(struct man *, int, int, enum mant);
 int		  man_head_alloc(struct man *, int, int, enum mant);
 int		  man_body_alloc(struct man *, int, int, enum mant);
 int		  man_elem_alloc(struct man *, int, int, enum mant);
-void		  man_node_free(struct man_node *);
-void		  man_node_freelist(struct man_node *);
-void		  man_node_unlink(struct man *, struct man_node *);
+void		  man_node_delete(struct man *, struct man_node *);
 void		  man_hash_init(void);
 enum	mant	  man_hash_find(const char *);
 int		  man_macroend(struct man *);
diff --git a/man.c b/man.c
index d09a17b6..5b72650c 100644
--- a/man.c
+++ b/man.c
@@ -1,4 +1,4 @@
-/*	$Id: man.c,v 1.53 2010/03/23 21:50:43 kristaps Exp $ */
+/*	$Id: man.c,v 1.54 2010/03/24 20:10:53 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -49,7 +49,8 @@ const	char *const __man_merrnames[WERRMAX] = {
 	"scope open on exit", /* WEXITSCOPE */
 	"no scope context", /* WNOSCOPE */
 	"literal context already open", /* WOLITERAL */
-	"no literal context open" /* WNLITERAL */
+	"no literal context open", /* WNLITERAL */
+	"invalid nesting of roff declarations", /* WROFFNEST */
 };
 
 const	char *const __man_macronames[MAN_MAX] = {		 
@@ -61,7 +62,9 @@ const	char *const __man_macronames[MAN_MAX] = {
 	"RI",		"na",		"i",		"sp",
 	"nf",		"fi",		"r",		"RE",
 	"RS",		"DT",		"UC",		"PD",
-	"Sp",		"Vb",		"Ve",
+	"Sp",		"Vb",		"Ve",		"de",
+	"dei",		"am",		"ami",		"ig",
+	".",
 	};
 
 const	char * const *man_macronames = __man_macronames;
@@ -70,6 +73,9 @@ static	struct man_node	*man_node_alloc(int, int,
 				enum man_type, enum mant);
 static	int		 man_node_append(struct man *, 
 				struct man_node *);
+static	void		 man_node_free(struct man_node *);
+static	void		 man_node_unlink(struct man *, 
+				struct man_node *);
 static	int		 man_ptext(struct man *, int, char *);
 static	int		 man_pmacro(struct man *, int, char *);
 static	void		 man_free1(struct man *);
@@ -160,7 +166,7 @@ man_free1(struct man *man)
 {
 
 	if (man->first)
-		man_node_freelist(man->first);
+		man_node_delete(man, man->first);
 	if (man->meta.title)
 		free(man->meta.title);
 	if (man->meta.source)
@@ -179,6 +185,7 @@ man_alloc1(struct man *m)
 	m->last = mandoc_calloc(1, sizeof(struct man_node));
 	m->first = m->last;
 	m->last->type = MAN_ROOT;
+	m->last->tok = MAN_MAX;
 	m->next = MAN_NEXT_CHILD;
 }
 
@@ -206,6 +213,7 @@ man_node_append(struct man *man, struct man_node *p)
 		/* NOTREACHED */
 	}
 	
+	assert(p->parent);
 	p->parent->nchild++;
 
 	if ( ! man_valid_pre(man, p))
@@ -336,30 +344,29 @@ man_word_alloc(struct man *m, int line, int pos, const char *word)
 }
 
 
-void
+/*
+ * Free all of the resources held by a node.  This does NOT unlink a
+ * node from its context; for that, see man_node_unlink().
+ */
+static void
 man_node_free(struct man_node *p)
 {
 
 	if (p->string)
 		free(p->string);
-	if (p->parent)
-		p->parent->nchild--;
 	free(p);
 }
 
 
 void
-man_node_freelist(struct man_node *p)
+man_node_delete(struct man *m, struct man_node *p)
 {
-	struct man_node	*n;
 
-	if (p->child)
-		man_node_freelist(p->child);
-	assert(0 == p->nchild);
-	n = p->next;
+	while (p->child)
+		man_node_delete(m, p->child);
+
+	man_node_unlink(m, p);
 	man_node_free(p);
-	if (n)
-		man_node_freelist(n);
 }
 
 
@@ -568,8 +575,7 @@ man_pmacro(struct man *m, int ln, char *buf)
 		if ( ! man_nwarn(m, n, WLNSCOPE))
 			return(0);
 
-		man_node_unlink(m, n);
-		man_node_free(n);
+		man_node_delete(m, n);
 		m->flags &= ~MAN_ELINE;
 	}
 
@@ -584,6 +590,9 @@ out:
 	/* 
 	 * We weren't in a block-line scope when entering the
 	 * above-parsed macro, so return.
+	 *
+	 * FIXME: this prohibits the nesting of blocks (e.g., `de' and
+	 * family) within BLINE or ELINE systems.  This is annoying.
 	 */
 
 	if ( ! (MAN_BLINE & fl)) {
@@ -672,26 +681,43 @@ man_err(struct man *m, int line, int pos, int iserr, enum merr type)
 }
 
 
-void
+/*
+ * Unlink a node from its context.  If "m" is provided, the last parse
+ * point will also be adjusted accordingly.
+ */
+static void
 man_node_unlink(struct man *m, struct man_node *n)
 {
 
-	if (n->prev) {
+	/* Adjust siblings. */
+
+	if (n->prev)
 		n->prev->next = n->next;
-		if (m->last == n) {
-			assert(NULL == n->next);
+	if (n->next)
+		n->next->prev = n->prev;
+
+	/* Adjust parent. */
+
+	if (n->parent) {
+		n->parent->nchild--;
+		if (n->parent->child == n)
+			n->parent->child = n->prev ? n->prev : n->next;
+	}
+
+	/* Adjust parse point, if applicable. */
+
+	if (m && m->last == n) {
+		/*XXX: this can occur when bailing from validation. */
+		/*assert(NULL == n->next);*/
+		if (n->prev) {
 			m->last = n->prev;
 			m->next = MAN_NEXT_SIBLING;
-		}
-	} else {
-		n->parent->child = n->next;
-		if (m->last == n) {
-			assert(NULL == n->next);
+		} else {
 			m->last = n->parent;
 			m->next = MAN_NEXT_CHILD;
 		}
 	}
 
-	if (n->next)
-		n->next->prev = n->prev;
+	if (m && m->first == n)
+		m->first = NULL;
 }
diff --git a/man.h b/man.h
index 9d4a7ef2..947e1efd 100644
--- a/man.h
+++ b/man.h
@@ -1,4 +1,4 @@
-/*	$Id: man.h,v 1.25 2010/03/23 21:50:43 kristaps Exp $ */
+/*	$Id: man.h,v 1.26 2010/03/24 20:10:53 kristaps Exp $ */
 /*
  * Copyright (c) 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -55,6 +55,12 @@ enum	mant {
 	MAN_Sp,
 	MAN_Vb,
 	MAN_Ve,
+	MAN_de,
+	MAN_dei,
+	MAN_am,
+	MAN_ami,
+	MAN_ig,
+	MAN_dot,
 	MAN_MAX,
 };
 
diff --git a/man_action.c b/man_action.c
index c4a8cd4c..53f79b3b 100644
--- a/man_action.c
+++ b/man_action.c
@@ -1,4 +1,4 @@
-/*	$Id: man_action.c,v 1.28 2010/03/24 03:46:02 kristaps Exp $ */
+/*	$Id: man_action.c,v 1.29 2010/03/24 20:10:53 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -30,6 +30,7 @@ struct	actions {
 };
 
 static	int	  post_TH(struct man *);
+static	int	  post_de(struct man *);
 static	int	  post_fi(struct man *);
 static	int	  post_nf(struct man *);
 
@@ -69,6 +70,12 @@ const	struct actions man_actions[MAN_MAX] = {
 	{ NULL }, /* Sp */
 	{ post_nf }, /* Vb */
 	{ post_fi }, /* Ve */
+	{ post_de }, /* de */
+	{ post_de }, /* dei */
+	{ post_de }, /* am */
+	{ post_de }, /* ami */
+	{ post_de }, /* ig */
+	{ NULL }, /* . */
 };
 
 
@@ -108,6 +115,21 @@ post_fi(struct man *m)
 
 
 static int
+post_de(struct man *m)
+{
+
+	/*
+	 * XXX: for the time being, we indiscriminately remove roff
+	 * instructions from the parse stream.
+	 */
+	if (MAN_BLOCK == m->last->type)
+		man_node_delete(m, m->last);
+
+	return(1);
+}
+
+
+static int
 post_nf(struct man *m)
 {
 
@@ -179,8 +201,10 @@ post_TH(struct man *m)
 	if (n && (n = n->next))
 		m->meta.vol = mandoc_strdup(n->string);
 
-	n = m->last;
-	man_node_unlink(m, n);
-	man_node_freelist(n);
+	/*
+	 * Remove the `TH' node after we've processed it for our
+	 * meta-data.
+	 */
+	man_node_delete(m, m->last);
 	return(1);
 }
diff --git a/man_html.c b/man_html.c
index a759bd9d..add096ca 100644
--- a/man_html.c
+++ b/man_html.c
@@ -1,4 +1,4 @@
-/*	$Id: man_html.c,v 1.29 2010/03/23 11:30:48 kristaps Exp $ */
+/*	$Id: man_html.c,v 1.30 2010/03/24 20:10:53 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -105,7 +105,13 @@ static	const struct htmlman mans[MAN_MAX] = {
 	{ man_ign_pre, NULL }, /* PD */
 	{ man_br_pre, NULL }, /* Sp */
 	{ man_ign_pre, NULL }, /* Vb */
-	{ NULL, NULL }, /* Vi */
+	{ NULL, NULL }, /* Ve */
+	{ man_ign_pre, NULL }, /* de */
+	{ man_ign_pre, NULL }, /* dei */
+	{ man_ign_pre, NULL }, /* am */
+	{ man_ign_pre, NULL }, /* ami */
+	{ man_ign_pre, NULL }, /* ig */
+	{ NULL, NULL }, /* . */
 };
 
 
diff --git a/man_macro.c b/man_macro.c
index 70c1ab55..5281a905 100644
--- a/man_macro.c
+++ b/man_macro.c
@@ -1,4 +1,4 @@
-/*	$Id: man_macro.c,v 1.34 2010/03/24 03:46:02 kristaps Exp $ */
+/*	$Id: man_macro.c,v 1.35 2010/03/24 20:10:53 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -31,9 +31,11 @@ enum	rew {
 	REW_HALT,
 };
 
-static	int		 in_line_eoln(MACRO_PROT_ARGS);
-static	int		 blk_imp(MACRO_PROT_ARGS);
 static	int		 blk_close(MACRO_PROT_ARGS);
+static	int		 blk_dotted(MACRO_PROT_ARGS);
+static	int		 blk_exp(MACRO_PROT_ARGS);
+static	int		 blk_imp(MACRO_PROT_ARGS);
+static	int		 in_line_eoln(MACRO_PROT_ARGS);
 
 static	int		 rew_scope(enum man_type, 
 				struct man *, enum mant);
@@ -71,13 +73,19 @@ const	struct man_macro __man_macros[MAN_MAX] = {
 	{ in_line_eoln, 0 }, /* fi */
 	{ in_line_eoln, 0 }, /* r */
 	{ blk_close, 0 }, /* RE */
-	{ blk_imp, MAN_EXPLICIT }, /* RS */
+	{ blk_exp, MAN_EXPLICIT }, /* RS */
 	{ in_line_eoln, 0 }, /* DT */
 	{ in_line_eoln, 0 }, /* UC */
 	{ in_line_eoln, 0 }, /* PD */
 	{ in_line_eoln, MAN_NSCOPED }, /* Sp */
 	{ in_line_eoln, 0 }, /* Vb */
 	{ in_line_eoln, 0 }, /* Ve */
+	{ blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* de */
+	{ blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* dei */
+	{ blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* am */
+	{ blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* ami */
+	{ blk_exp, MAN_EXPLICIT | MAN_NOCLOSE}, /* ig */
+	{ blk_dotted, 0 }, /* . */
 };
 
 const	struct man_macro * const man_macros = __man_macros;
@@ -88,7 +96,6 @@ man_unscope(struct man *m, const struct man_node *n)
 {
 
 	assert(n);
-	m->next = MAN_NEXT_SIBLING;
 
 	/* LINTED */
 	while (m->last != n) {
@@ -102,7 +109,13 @@ man_unscope(struct man *m, const struct man_node *n)
 
 	if ( ! man_valid_post(m))
 		return(0);
-	return(man_action_post(m));
+	if ( ! man_action_post(m))
+		return(0);
+
+	m->next = MAN_ROOT == m->last->type ? 
+		MAN_NEXT_CHILD : MAN_NEXT_SIBLING;
+
+	return(1);
 }
 
 
@@ -206,6 +219,37 @@ rew_scope(enum man_type type, struct man *m, enum mant tok)
 
 /* ARGSUSED */
 int
+blk_dotted(MACRO_PROT_ARGS)
+{
+	enum mant	 ntok;
+	struct man_node	*nn;
+
+	for (nn = m->last->parent; nn; nn = nn->parent)
+		if (nn->tok == MAN_de || nn->tok == MAN_dei ||
+				nn->tok == MAN_am ||
+				nn->tok == MAN_ami ||
+				nn->tok == MAN_ig) {
+			ntok = nn->tok;
+			break;
+		}
+
+	if (NULL == nn) {
+		if ( ! man_pwarn(m, line, ppos, WNOSCOPE))
+			return(0);
+		return(1);
+	}
+
+	if ( ! rew_scope(MAN_BODY, m, ntok))
+		return(0);
+	if ( ! rew_scope(MAN_BLOCK, m, ntok))
+		return(0);
+
+	return(1);
+}
+
+
+/* ARGSUSED */
+int
 blk_close(MACRO_PROT_ARGS)
 {
 	enum mant	 	 ntok;
@@ -232,11 +276,61 @@ blk_close(MACRO_PROT_ARGS)
 		return(0);
 	if ( ! rew_scope(MAN_BLOCK, m, ntok))
 		return(0);
-	m->next = MAN_NEXT_SIBLING;
+
 	return(1);
 }
 
 
+int
+blk_exp(MACRO_PROT_ARGS)
+{
+	int		 w, la;
+	char		*p;
+	struct man_node	*n;
+
+	/* 
+	 * Close out prior scopes.  "Regular" explicit macros cannot be
+	 * nested, but we allow roff macros to be placed just about
+	 * anywhere.
+	 */
+
+	if ( ! (MAN_NOCLOSE & man_macros[tok].flags)) {
+		if ( ! rew_scope(MAN_BODY, m, tok))
+			return(0);
+		if ( ! rew_scope(MAN_BLOCK, m, tok))
+			return(0);
+	}
+
+	if ( ! man_block_alloc(m, line, ppos, tok))
+		return(0);
+	if ( ! man_head_alloc(m, line, ppos, tok))
+		return(0);
+
+	n = m->last;
+
+	for (;;) {
+		la = *pos;
+		w = man_args(m, line, pos, buf, &p);
+
+		if (-1 == w)
+			return(0);
+		if (0 == w)
+			break;
+
+		if ( ! man_word_alloc(m, line, la, p))
+			return(0);
+	}
+
+	assert(m);
+	assert(tok != MAN_MAX);
+
+	if ( ! rew_scope(MAN_HEAD, m, tok))
+		return(0);
+	return(man_body_alloc(m, line, ppos, tok));
+}
+
+
+
 /*
  * Parse an implicit-block macro.  These contain a MAN_HEAD and a
  * MAN_BODY contained within a MAN_BLOCK.  Rules for closing out other
@@ -296,7 +390,6 @@ blk_imp(MACRO_PROT_ARGS)
 
 	if ( ! rew_scope(MAN_HEAD, m, tok))
 		return(0);
-
 	return(man_body_alloc(m, line, ppos, tok));
 }
 
@@ -371,8 +464,9 @@ in_line_eoln(MACRO_PROT_ARGS)
 		return(0);
 	if (m->last->type != MAN_ROOT && ! man_action_post(m))
 		return(0);
-	if (m->last->type != MAN_ROOT)
-		m->next = MAN_NEXT_SIBLING;
+
+	m->next = MAN_ROOT == m->last->type ?
+		MAN_NEXT_CHILD : MAN_NEXT_SIBLING;
 
 	return(1);
 }
diff --git a/man_term.c b/man_term.c
index 2e7397e0..ee833bdc 100644
--- a/man_term.c
+++ b/man_term.c
@@ -1,4 +1,4 @@
-/*	$Id: man_term.c,v 1.58 2010/03/23 12:42:22 kristaps Exp $ */
+/*	$Id: man_term.c,v 1.59 2010/03/24 20:10:53 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -142,6 +142,12 @@ static	const struct termact termacts[MAN_MAX] = {
  	{ pre_sp, NULL, MAN_NOTEXT }, /* Sp */
  	{ pre_nf, NULL, 0 }, /* Vb */
  	{ pre_fi, NULL, 0 }, /* Ve */
+ 	{ pre_ign, NULL, MAN_NOTEXT }, /* de */
+ 	{ pre_ign, NULL, MAN_NOTEXT }, /* dei */
+ 	{ pre_ign, NULL, MAN_NOTEXT }, /* am */
+ 	{ pre_ign, NULL, MAN_NOTEXT }, /* ami */
+ 	{ pre_ign, NULL, MAN_NOTEXT }, /* ig */
+ 	{ NULL, NULL, 0 }, /* . */
 };
 
 
@@ -785,6 +791,8 @@ post_RS(DECL_ARGS)
 	case (MAN_BLOCK):
 		mt->offset = mt->lmargin = INDENT;
 		break;
+	case (MAN_HEAD):
+		break;
 	default:
 		term_newln(p);
 		p->offset = INDENT;
diff --git a/man_validate.c b/man_validate.c
index f0b8164f..db3b1d0e 100644
--- a/man_validate.c
+++ b/man_validate.c
@@ -1,4 +1,4 @@
-/*	$Id: man_validate.c,v 1.30 2010/03/23 11:30:48 kristaps Exp $ */
+/*	$Id: man_validate.c,v 1.31 2010/03/24 20:10:53 kristaps Exp $ */
 /*
  * Copyright (c) 2008, 2009 Kristaps Dzonsons <kristaps@kth.se>
  *
@@ -46,6 +46,7 @@ static	int	  check_ge2(CHKARGS);
 static	int	  check_le5(CHKARGS);
 static	int	  check_par(CHKARGS);
 static	int	  check_part(CHKARGS);
+static	int	  check_roff(CHKARGS);
 static	int	  check_root(CHKARGS);
 static	int	  check_sec(CHKARGS);
 static	int	  check_text(CHKARGS);
@@ -57,6 +58,7 @@ static	v_check	  posts_part[] = { check_part, NULL };
 static	v_check	  posts_sec[] = { check_sec, NULL };
 static	v_check	  posts_le1[] = { check_le1, NULL };
 static	v_check	  pres_bline[] = { check_bline, NULL };
+static	v_check	  pres_roff[] = { check_bline, check_roff, NULL };
 
 static	const struct man_valid man_valids[MAN_MAX] = {
 	{ NULL, posts_eq0 }, /* br */
@@ -94,6 +96,12 @@ static	const struct man_valid man_valids[MAN_MAX] = {
 	{ NULL, posts_eq0 }, /* Sp */
 	{ pres_bline, posts_le1 }, /* Vb */
 	{ pres_bline, posts_eq0 }, /* Ve */
+	{ pres_roff, NULL }, /* de */
+	{ pres_roff, NULL }, /* dei */
+	{ pres_roff, NULL }, /* am */
+	{ pres_roff, NULL }, /* ami */
+	{ pres_roff, NULL }, /* ig */
+	{ NULL, NULL }, /* . */
 };
 
 
@@ -284,6 +292,24 @@ check_bline(CHKARGS)
 	assert( ! (MAN_ELINE & m->flags));
 	if (MAN_BLINE & m->flags)
 		return(man_nerr(m, n, WLNSCOPE));
+
 	return(1);
 }
 
+
+static int
+check_roff(CHKARGS)
+{
+
+	if (MAN_BLOCK != n->type)
+		return(1);
+
+	for (n = n->parent; n; n = n->parent)
+		if (MAN_de == n->tok || MAN_dei == n->tok ||
+				MAN_am == n->tok || 
+				MAN_ami == n->tok ||
+				MAN_ig == n->tok)
+			return(man_nerr(m, n, WROFFNEST));
+
+	return(1);
+}