From 72a53430c7b72293f68d675a5a1c1848d4d682dd Mon Sep 17 00:00:00 2001
From: pstef <pstef@FreeBSD.org>
Date: Sun, 3 Jun 2018 16:21:15 +0000
Subject: [PATCH] indent(1): disjoint parser state from lexi()

The function is sometimes used as a look-ahead, so ideally it should bear
no information about parser state.
---
 usr.bin/indent/indent.c              |  3 ++
 usr.bin/indent/indent_codes.h        |  2 +-
 usr.bin/indent/lexi.c                | 42 +++++++++-------------------
 usr.bin/indent/tests/struct.0        |  8 ++++++
 usr.bin/indent/tests/struct.0.stdout | 10 +++++++
 5 files changed, 35 insertions(+), 30 deletions(-)

diff --git a/usr.bin/indent/indent.c b/usr.bin/indent/indent.c
index d19d2c100dfd..1d55435fbaf1 100644
--- a/usr.bin/indent/indent.c
+++ b/usr.bin/indent/indent.c
@@ -995,6 +995,9 @@ main(int argc, char **argv)
 	    prefix_blankline_requested = 0;
 	    goto copy_id;
 
+	case structure:
+	    if (ps.p_l_follow > 0)
+		goto copy_id;
 	case decl:		/* we have a declaration type (int, etc.) */
 	    parse(decl);	/* let parser worry about indentation */
 	    if (ps.last_token == rparen && ps.tos <= 1) {
diff --git a/usr.bin/indent/indent_codes.h b/usr.bin/indent/indent_codes.h
index 5e7c37f4af35..0d043279c4e1 100644
--- a/usr.bin/indent/indent_codes.h
+++ b/usr.bin/indent/indent_codes.h
@@ -74,4 +74,4 @@
 #define storage		34
 #define funcname	35
 #define type_def	36
-
+#define structure	37
diff --git a/usr.bin/indent/lexi.c b/usr.bin/indent/lexi.c
index ef76776f2ca4..1cc6ff19e2e6 100644
--- a/usr.bin/indent/lexi.c
+++ b/usr.bin/indent/lexi.c
@@ -145,8 +145,6 @@ lexi(struct parser_state *state)
 {
     int         unary_delim;	/* this is set to 1 if the current token
 				 * forces a following operator to be unary */
-    static int  last_code;	/* the last token type returned */
-    static int  l_struct;	/* set to 1 if the last token was 'struct' */
     int         code;		/* internal code to be returned */
     char        qchar;		/* the delimiter character for a string */
 
@@ -283,21 +281,17 @@ lexi(struct parser_state *state)
 		fill_buffer();
 	}
 	state->keyword = 0;
-	if (l_struct && !state->p_l_follow) {
+	if (state->last_token == structure && !state->p_l_follow) {
 				/* if last token was 'struct' and we're not
 				 * in parentheses, then this token
 				 * should be treated as a declaration */
-	    l_struct = false;
-	    last_code = ident;
 	    state->last_u_d = true;
 	    return (decl);
 	}
-	state->last_u_d = l_struct;	/* Operator after identifier is
-					 * binary unless last token was
-					 * 'struct' */
-	l_struct = false;
-	last_code = ident;	/* Remember that this is the code we will
-				 * return */
+	/*
+	 * Operator after identifier is binary unless last token was 'struct'
+	 */
+	state->last_u_d = (state->last_token == structure);
 
 	p = bsearch(s_token,
 	    specials,
@@ -326,21 +320,17 @@ lexi(struct parser_state *state)
 		return (casestmt);
 
 	    case 3:		/* a "struct" */
-		/*
-		 * Next time around, we will want to know that we have had a
-		 * 'struct'
-		 */
-		l_struct = true;
 		/* FALLTHROUGH */
-
 	    case 4:		/* one of the declaration keywords */
 	    found_typename:
 		if (state->p_l_follow) {
 		    /* inside parens: cast, param list, offsetof or sizeof */
 		    state->cast_mask |= (1 << state->p_l_follow) & ~state->not_cast_mask;
-		    break;
 		}
-		last_code = decl;
+		if (p != NULL && p->rwcode == 3)
+		    return (structure);
+		if (state->p_l_follow)
+		    break;
 		return (decl);
 
 	    case 5:		/* if, while, for */
@@ -369,7 +359,7 @@ lexi(struct parser_state *state)
 	    strncpy(state->procname, token, sizeof state->procname - 1);
 	    if (state->in_decl)
 		state->in_parameter_declaration = 1;
-	    return (last_code = funcname);
+	    return (funcname);
     not_proc:;
 	}
 	/*
@@ -385,13 +375,11 @@ lexi(struct parser_state *state)
 		    state->last_token == lbrace || state->last_token == rbrace)) {
 	    state->keyword = 4;	/* a type name */
 	    state->last_u_d = true;
-	    last_code = decl;
 	    return decl;
 	}
-	if (last_code == decl)	/* if this is a declared variable, then
-				 * following sign is unary */
+	if (state->last_token == decl)	/* if this is a declared variable,
+					 * then following sign is unary */
 	    state->last_u_d = true;	/* will make "int a -1" work */
-	last_code = ident;
 	return (ident);		/* the ident is not in the list */
     }				/* end of procesing for alpanum character */
 
@@ -536,7 +524,7 @@ lexi(struct parser_state *state)
 	    /* check for doubled character */
 	    *e_token++ = *buf_ptr++;
 	    /* buffer overflow will be checked at end of loop */
-	    if (last_code == ident || last_code == rparen) {
+	    if (state->last_token == ident || state->last_token == rparen) {
 		code = (state->last_u_d ? unary_op : postop);
 		/* check for following ++ or -- */
 		unary_delim = false;
@@ -617,10 +605,6 @@ lexi(struct parser_state *state)
 
 
     }				/* end of switch */
-    if (code != newline) {
-	l_struct = false;
-	last_code = code;
-    }
     if (buf_ptr >= buf_end)	/* check for input buffer empty */
 	fill_buffer();
     state->last_u_d = unary_delim;
diff --git a/usr.bin/indent/tests/struct.0 b/usr.bin/indent/tests/struct.0
index 9dfc31de9a03..83142bfb1972 100644
--- a/usr.bin/indent/tests/struct.0
+++ b/usr.bin/indent/tests/struct.0
@@ -1,4 +1,7 @@
 /* $FreeBSD$ */
+
+int f(struct x *a);
+
 /* See r303485 */
 void
 t(void)
@@ -11,3 +14,8 @@ t(void)
 		{ F, G }
 	};
 }
+
+void u(struct x a) {
+	int b;
+	struct y c = (struct y *)&a;
+}
diff --git a/usr.bin/indent/tests/struct.0.stdout b/usr.bin/indent/tests/struct.0.stdout
index ef62a12e8f8e..38613128654f 100644
--- a/usr.bin/indent/tests/struct.0.stdout
+++ b/usr.bin/indent/tests/struct.0.stdout
@@ -1,4 +1,7 @@
 /* $FreeBSD$ */
+
+int		f(struct x *a);
+
 /* See r303485 */
 void
 t(void)
@@ -11,3 +14,10 @@ t(void)
 		{F, G}
 	};
 }
+
+void
+u(struct x a)
+{
+	int		b;
+	struct y	c = (struct y *)&a;
+}