From 7c77bd90e487a5e68a8f1d44c232a6d605490504 Mon Sep 17 00:00:00 2001
From: Nick Gasson <nick@nickg.me.uk>
Date: Sat, 5 Nov 2016 19:31:27 +0000
Subject: [PATCH] Fix parsing of ASCII special characters

---
 src/common.c           |  3 +++
 test/parse/literal.vhd |  2 ++
 test/test_parse.c      | 11 ++++++++++-
 3 files changed, 15 insertions(+), 1 deletion(-)

diff --git a/src/common.c b/src/common.c
index 537c3cd8..2379015e 100644
--- a/src/common.c
+++ b/src/common.c
@@ -721,6 +721,9 @@ tree_t str_to_literal(const char *start, const char *end, type_t type)
    }
 
    for (const char *p = start; *p != '\0' && p != end; p++) {
+      if (*(const unsigned char *)p == 0x81)
+         continue;   // Allow UTF-8 encoded ASCII characters
+
       const char ch[] = { '\'', *p, '\'', '\0' };
       ident_t id = ident_new(ch);
       tree_t ref = tree_new(T_REF);
diff --git a/test/parse/literal.vhd b/test/parse/literal.vhd
index 337a645b..df85ac89 100644
--- a/test/parse/literal.vhd
+++ b/test/parse/literal.vhd
@@ -1,3 +1,4 @@
+-- -*- coding: latin-1 -*-
 ARCHITECTURE a OF e IS
   SIGNAL pos : INTEGER := 64;
   SIGNAL neg : INTEGER := -265;
@@ -22,6 +23,7 @@ ARCHITECTURE a OF e IS
   CONSTANT s : STRING  := % %;
   CONSTANT t : STRING  := %A%;
   CONSTANT u : STRING  := %%%%;
+  constant v : string  := "©";
 BEGIN
 
 END ARCHITECTURE;
diff --git a/test/test_parse.c b/test/test_parse.c
index 45d7f821..931b3c5a 100644
--- a/test/test_parse.c
+++ b/test/test_parse.c
@@ -776,7 +776,7 @@ START_TEST(test_literal)
    a = parse();
    fail_if(a == NULL);
    fail_unless(tree_kind(a) == T_ARCH);
-   fail_unless(tree_decls(a) == 23);
+   fail_unless(tree_decls(a) == 24);
 
    d = tree_decl(a, 0);
    fail_unless(tree_ident(d) == ident_new("POS"));
@@ -1018,6 +1018,15 @@ START_TEST(test_literal)
    fail_unless(tree_chars(v) == 1);
    fail_unless(tree_ident(tree_char(v, 0)) == ident_new("'%'"));
 
+   d = tree_decl(a, 23);
+   fail_unless(tree_kind(d) == T_CONST_DECL);
+   fail_unless(tree_ident(d) == ident_new("V"));
+   v = tree_value(d);
+   fail_unless(tree_kind(v) == T_LITERAL);
+   fail_unless(tree_subkind(v) == L_STRING);
+   fail_unless(tree_chars(v) == 1);
+   fail_unless(ident_char(tree_ident(tree_char(v, 0)), 1) == (char)0xa9);
+
    a = parse();
    fail_unless(a == NULL);
 
-- 
2.39.2