Escape-sequence validation in place (for nodes).

11 files changed, 220 insertions, 24 deletions

diff --git a/Makefile b/Makefile
index 443c6e99..29d51539 100644
--- a/Makefile
+++ b/Makefile
@@ -78,9 +78,11 @@ FAIL	= regress/test.empty \
 	  regress/test.prologue.31 \
 	  regress/test.prologue.32 \
 	  regress/test.prologue.33 \
-	  regress/test.sh.01 \
-	  regress/test.sh.02 \
-	  regress/test.sh.03
+	  regress/test.sh.03 \
+	  regress/test.escape.01 \
+	  regress/test.escape.02 \
+	  regress/test.escape.03 \
+	  regress/test.escape.04
 
 SUCCEED	= regress/test.prologue.05 \
 	  regress/test.prologue.07 \
@@ -103,7 +105,10 @@ SUCCEED	= regress/test.prologue.05 \
 	  regress/test.list.03 \
 	  regress/test.list.04 \
 	  regress/test.list.05 \
-	  regress/test.list.06
+	  regress/test.list.06 \
+	  regress/test.sh.01 \
+	  regress/test.sh.02 \
+	  regress/test.escape.00
 
 REGRESS	= $(FAIL) $(SUCCEED)
 
diff --git a/argv.c b/argv.c
index c7e8f46a..51ef35b2 100644
--- a/argv.c
+++ b/argv.c
@@ -35,7 +35,7 @@
 #define	ARGS_DELIM	(1 << 1)
 #define	ARGS_TABSEP	(1 << 2)
 
-static	int		 lookup(int, const char *);
+static	int		 argv_a2arg(int, const char *);
 static	int		 args(struct mdoc *, int, int *, 
 				char *, int, char **);
 static	int		 argv(struct mdoc *, int,
@@ -466,7 +466,7 @@ args(struct mdoc *mdoc, int line,
 
 
 static int
-lookup(int tok, const char *argv)
+argv_a2arg(int tok, const char *argv)
 {
 
 	switch (tok) {
@@ -747,7 +747,7 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok,
 	if (buf[*pos])
 		buf[(*pos)++] = 0;
 
-	if (MDOC_ARG_MAX == (v->arg = lookup(tok, p))) {
+	if (MDOC_ARG_MAX == (v->arg = argv_a2arg(tok, p))) {
 		if ( ! pwarn(mdoc, line, i, WARGVPARM))
 			return(ARGV_ERROR);
 		return(ARGV_WORD);
diff --git a/private.h b/private.h
index 89ac19d4..e6e8210c 100644
--- a/private.h
+++ b/private.h
@@ -125,6 +125,7 @@ void		 *mdoc_tokhash_alloc(void);
 int		  mdoc_tokhash_find(const void *, const char *);
 void		  mdoc_tokhash_free(void *);
 int		  mdoc_iscdelim(char);
+size_t		  mdoc_isescape(const char *);
 enum	mdoc_sec  mdoc_atosec(const char *);
 enum	mdoc_msec mdoc_atomsec(const char *);
 enum	mdoc_vol  mdoc_atovol(const char *);
diff --git a/regress/test.escape.00 b/regress/test.escape.00
new file mode 100644
index 00000000..28df79e5
--- /dev/null
+++ b/regress/test.escape.00
@@ -0,0 +1,20 @@
+.\" 
+.Dd $Mdocdate$
+.Dt mdoc 3
+.Os
+.\"
+.Sh NAME
+.Nm mdoc_free
+.Nd mdoc macro compiler library
+.\"
+.Sh SYNOPSIS
+Valid escape: \(ab 
+Valid escape: \[d]
+Valid escape: \[dsdfajsdflaksjfhalksjdfh__----]
+Valid escape: \\
+Valid escape: \e
+Valid escape: \`
+Valid escape: \'
+Valid escape: \.
+Valid escape: \-
+Valid escape: \ 
diff --git a/regress/test.escape.01 b/regress/test.escape.01
new file mode 100644
index 00000000..2bf37cf6
--- /dev/null
+++ b/regress/test.escape.01
@@ -0,0 +1,20 @@
+.\" 
+.Dd $Mdocdate$
+.Dt mdoc 3
+.Os
+.\"
+.Sh NAME
+.Nm mdoc_free
+.Nd mdoc macro compiler library
+.\"
+.Sh SYNOPSIS
+Valid escape: \(ab 
+Valid escape: \[d]
+Valid escape: \[dsdfajsdflaksjfhalksjdfh__----]
+Valid escape: \\
+Valid escape: \e
+Valid escape: \`
+Valid escape: \'
+Valid escape: \.
+Valid escape: \-
+Invalid escape: \"
diff --git a/regress/test.escape.02 b/regress/test.escape.02
new file mode 100644
index 00000000..1c727a8f
--- /dev/null
+++ b/regress/test.escape.02
@@ -0,0 +1,20 @@
+.\" 
+.Dd $Mdocdate$
+.Dt mdoc 3
+.Os
+.\"
+.Sh NAME
+.Nm mdoc_free
+.Nd mdoc macro compiler library
+.\"
+.Sh SYNOPSIS
+Valid escape: \(ab 
+Valid escape: \[d]
+Valid escape: \[dsdfajsdflaksjfhalksjdfh__----]
+Valid escape: \\
+Valid escape: \e
+Valid escape: \`
+Valid escape: \'
+Valid escape: \.
+Valid escape: \-
+Invalid escape: \(
diff --git a/regress/test.escape.03 b/regress/test.escape.03
new file mode 100644
index 00000000..7fa92ecd
--- /dev/null
+++ b/regress/test.escape.03
@@ -0,0 +1,20 @@
+.\" 
+.Dd $Mdocdate$
+.Dt mdoc 3
+.Os
+.\"
+.Sh NAME
+.Nm mdoc_free
+.Nd mdoc macro compiler library
+.\"
+.Sh SYNOPSIS
+Valid escape: \(ab 
+Valid escape: \[d]
+Valid escape: \[dsdfajsdflaksjfhalksjdfh__----]
+Valid escape: \\
+Valid escape: \e
+Valid escape: \`
+Valid escape: \'
+Valid escape: \.
+Valid escape: \-
+Invalid escape: \
diff --git a/regress/test.escape.04 b/regress/test.escape.04
new file mode 100644
index 00000000..a393ab83
--- /dev/null
+++ b/regress/test.escape.04
@@ -0,0 +1,20 @@
+.\" 
+.Dd $Mdocdate$
+.Dt mdoc 3
+.Os
+.\"
+.Sh NAME
+.Nm mdoc_free
+.Nd mdoc macro compiler library
+.\"
+.Sh SYNOPSIS
+Valid escape: \(ab 
+Valid escape: \[d]
+Valid escape: \[dsdfajsdflaksjfhalksjdfh__----]
+Valid escape: \\
+Valid escape: \e
+Valid escape: \`
+Valid escape: \'
+Valid escape: \.
+Valid escape: \-
+Invalid escape: \[
diff --git a/regress/test.list.05 b/regress/test.list.05
index 18540423..08954e84 100644
--- a/regress/test.list.05
+++ b/regress/test.list.05
@@ -6,5 +6,5 @@
 .Nd description
 .Sh DESCRIPTION
 .Bl -column "xxxxx" "xxxxx"
-.It d \	 e	f
+.It d \ e	f
 .El
diff --git a/strings.c b/strings.c
index 2b7cd4fc..3e9415c5 100644
--- a/strings.c
+++ b/strings.c
@@ -35,6 +35,50 @@
 extern	char		*strptime(const char *, const char *, struct tm *);
 #endif
 
+
+size_t
+mdoc_isescape(const char *p)
+{
+	size_t		 c;
+	
+	if ('\\' != *p++)
+		return(0);
+
+	switch (*p) {
+	case ('\\'):
+		/* FALLTHROUGH */
+	case ('\''):
+		/* FALLTHROUGH */
+	case ('`'):
+		/* FALLTHROUGH */
+	case ('-'):
+		/* FALLTHROUGH */
+	case (' '):
+		/* FALLTHROUGH */
+	case ('.'):
+		/* FALLTHROUGH */
+	case ('e'):
+		return(2);
+	case ('('):
+		if (0 == *++p)
+			return(0);
+		if (0 == *++p)
+			return(0);
+		return(4);
+	case ('['):
+		break;
+	default:
+		return(0);
+	}
+
+	for (c = 3, p++; *p && ']' != *p; p++, c++)
+		if (isspace(*p))
+			break;
+
+	return(*p == ']' ? c : 0);
+}
+
+
 int
 mdoc_iscdelim(char p)
 {
diff --git a/validate.c b/validate.c
index 871d913e..cafcd7c0 100644
--- a/validate.c
+++ b/validate.c
@@ -35,7 +35,6 @@ typedef	int	(*v_post)(struct mdoc *);
 /* FIXME: some sections should only occur in specific msecs. */
 /* FIXME: ignoring Pp. */
 /* FIXME: math symbols. */
-/* FIXME: valid character-escape checks. */
 /* FIXME: .Fd only in synopsis section. */
 
 struct	valids {
@@ -109,6 +108,7 @@ static	int	post_xr(struct mdoc *);
 static	int	post_nm(struct mdoc *);
 static	int	post_bf(struct mdoc *);
 static	int	post_root(struct mdoc *);
+static	int	pre_text(struct mdoc *, const struct mdoc_node *);
 
 /* Collections of pre-child-parse routines. */
 
@@ -386,6 +386,22 @@ check_msec(struct mdoc *mdoc, struct mdoc_node *node,
 
 
 static int
+check_parent(struct mdoc *mdoc, struct mdoc_node *n, 
+		int tok, enum mdoc_type t)
+{
+
+	assert(n->parent);
+	if ((MDOC_ROOT == t || tok == n->parent->tok) &&
+			(t == n->parent->type))
+		return(1);
+
+	return(mdoc_nerr(mdoc, n, "require parent %s",
+		MDOC_ROOT == t ? "<root>" : mdoc_macronames[tok]));
+}
+
+
+
+static int
 pre_display(struct mdoc *mdoc, struct mdoc_node *node)
 {
 	struct mdoc_node *n;
@@ -922,6 +938,26 @@ ebool(struct mdoc *mdoc)
 
 
 static int
+pre_text(struct mdoc *mdoc, const struct mdoc_node *n)
+{
+	size_t		 c;
+	const char	*p;
+
+	for (p = n->data.text.string; *p; p++) {
+		if ('\\' != *p)
+			continue;
+		if ((c = mdoc_isescape(p))) {
+			p += (c - 1);
+			continue;
+		}
+		return(mdoc_nerr(mdoc, n, "bad escape sequence"));
+	}
+
+	return(1);
+}
+
+
+static int
 post_root(struct mdoc *mdoc)
 {
 
@@ -931,10 +967,10 @@ post_root(struct mdoc *mdoc)
 		return(mdoc_err(mdoc, "document lacks prologue"));
 
 	if (MDOC_BLOCK != mdoc->first->child->type)
-		return(mdoc_err(mdoc, "lacking post-prologue `%s'", 
+		return(mdoc_err(mdoc, "lacking post-prologue %s", 
 					mdoc_macronames[MDOC_Sh]));
 	if (MDOC_Sh != mdoc->first->child->tok)
-		return(mdoc_err(mdoc, "lacking post-prologue `%s'", 
+		return(mdoc_err(mdoc, "lacking post-prologue %s", 
 					mdoc_macronames[MDOC_Sh]));
 
 	return(1);
@@ -969,8 +1005,8 @@ post_sh_body(struct mdoc *mdoc)
 	 */
 
 	if (NULL == (n = mdoc->last->child))
-		return(mdoc_warn(mdoc, WARN_COMPAT, "section NAME "
-					"should contain %s and %s", 
+		return(mdoc_warn(mdoc, WARN_SYNTAX, 
+					"section should have %s and %s",
 					mdoc_macronames[MDOC_Nm],
 					mdoc_macronames[MDOC_Nd]));
 
@@ -979,9 +1015,8 @@ post_sh_body(struct mdoc *mdoc)
 			continue;
 		if (MDOC_TEXT == n->type)
 			continue;
-		if ( ! (mdoc_nwarn(mdoc, n, WARN_COMPAT, "section "
-					"NAME should contain %s as "
-					"initial body child", 
+		if ( ! (mdoc_nwarn(mdoc, n, WARN_SYNTAX, 
+					"section should have %s first",
 					mdoc_macronames[MDOC_Nm])))
 			return(0);
 	}
@@ -989,8 +1024,8 @@ post_sh_body(struct mdoc *mdoc)
 	if (MDOC_ELEM == n->type && MDOC_Nd == n->tok)
 		return(1);
 
-	return(mdoc_warn(mdoc, WARN_COMPAT, "section NAME should "
-				"contain %s as the last child",
+	return(mdoc_warn(mdoc, WARN_SYNTAX, 
+				"section should have %s last",
 				mdoc_macronames[MDOC_Nd]));
 }
 
@@ -1003,19 +1038,22 @@ post_sh_head(struct mdoc *mdoc)
 
 	assert(MDOC_Sh == mdoc->last->tok);
 
-	if ( ! xstrlcats(buf, mdoc->last->child, 64))
-		return(mdoc_err(mdoc, "macro parameters too long"));
+	if ( ! xstrlcats(buf, mdoc->last->child, sizeof(buf)))
+		return(mdoc_err(mdoc, "argument too long"));
 
 	sec = mdoc_atosec(buf);
 
 	if (SEC_BODY == mdoc->lastnamed && SEC_NAME != sec)
-		return(mdoc_err(mdoc, "section NAME must be first"));
+		return(mdoc_warn(mdoc, WARN_SYNTAX, 
+				"section NAME should be first"));
 	if (SEC_CUSTOM == sec)
 		return(1);
 	if (sec == mdoc->lastnamed)
-		return(mdoc_warn(mdoc, WARN_SYNTAX, "section repeated"));
+		return(mdoc_warn(mdoc, WARN_SYNTAX, 
+				"section repeated"));
 	if (sec < mdoc->lastnamed)
-		return(mdoc_warn(mdoc, WARN_SYNTAX, "section out of conventional order"));
+		return(mdoc_warn(mdoc, WARN_SYNTAX, 
+				"section out of order"));
 
 	return(1);
 }
@@ -1027,7 +1065,7 @@ mdoc_valid_pre(struct mdoc *mdoc, struct mdoc_node *node)
 	v_pre		*p;
 
 	if (MDOC_TEXT == node->type)
-		return(1);
+		return(pre_text(mdoc, node));
 	assert(MDOC_ROOT != node->type);
 
 	if (NULL == mdoc_valids[node->tok].pre)
@@ -1044,6 +1082,14 @@ mdoc_valid_post(struct mdoc *mdoc)
 {
 	v_post		*p;
 
+	/*
+	 * This check occurs after the macro's children have been filled
+	 * in: postfix validation.  Since this happens when we're
+	 * rewinding the scope tree, it's possible to have multiple
+	 * invocations (as by design, for now), we set bit MDOC_VALID to
+	 * indicate that we've validated.
+	 */
+
 	if (MDOC_VALID & mdoc->last->flags)
 		return(1);
 	mdoc->last->flags |= MDOC_VALID;