From 72c72010e12c96ee881dc5034913d6410af8f90b Mon Sep 17 00:00:00 2001
From: Christine Caulfield <ccaulfie@redhat.com>
Date: Thu, 5 Mar 2020 09:41:52 +0000
Subject: [PATCH 1/3] trie: Don't assume that chars are unsigned < 126

Trie fails on systems with unsigned chars when using characters over
126.

This needs looking over and running through the CI systems
---
 lib/trie.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/lib/trie.c b/lib/trie.c
index 0e5dd6e4b..5b1f8b019 100644
--- a/lib/trie.c
+++ b/lib/trie.c
@@ -64,8 +64,8 @@ static void trie_destroy_node(struct trie_node *node);
  * characters are stored in reverse to make accessing the
  * more common case (non-control chars) more space efficient.
  */
-#define TRIE_CHAR2INDEX(ch) (126 - ch)
-#define TRIE_INDEX2CHAR(idx) (126 - idx)
+#define TRIE_CHAR2INDEX(ch) abs(126 - ch)
+#define TRIE_INDEX2CHAR(idx) abs(126 - idx)
 
 
 static int32_t

From aa80a88615c0a0e56a489b009601b5d18eed6c8d Mon Sep 17 00:00:00 2001
From: Christine Caulfield <ccaulfie@redhat.com>
Date: Thu, 5 Mar 2020 13:44:17 +0000
Subject: [PATCH 2/3] test: Unit tests for high-characters in a map

---
 tests/check_map.c | 81 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 81 insertions(+)

diff --git a/tests/check_map.c b/tests/check_map.c
index e73c832ec..578e19bdf 100644
--- a/tests/check_map.c
+++ b/tests/check_map.c
@@ -41,6 +41,19 @@ const char *chars2[] = {
 	NULL,
 };
 
+const char *composers[] = {
+	"Béla Bartók",
+	"Zoltán Kodály",
+	"Ludwig van Beethoven",
+	"Wolfgang Amadeus Mozart",
+	"Leoš Janáček",
+	"Benjamin Britten",
+	"Josef Haydn",
+	"Claude Debussy",
+	"Charles Ives"
+};
+
+
 static char *notified_key = NULL;
 static void *notified_value = NULL;
 static void *notified_new_value = NULL;
@@ -717,6 +730,45 @@ test_map_load(qb_map_t *m, const char* test_name)
 	qb_log(LOG_INFO, "%25s %12.2f dels/sec (%d/%fs)\n", test_name, ops, count2, secs);
 }
 
+static void test_accents_load(qb_map_t *m, const char* test_name)
+{
+	int i;
+	int32_t res = 0;
+	int32_t count = 0;
+	int32_t count2;
+	void *value;
+
+	ck_assert(m != NULL);
+	/*
+	 * Load accented names
+	 */
+	for (i=0; i<sizeof(composers)/sizeof(char*); i++) {
+		qb_map_put(m, strdup(composers[i]), strdup(composers[i]));
+		count++;
+	}
+	ck_assert_int_eq(qb_map_count_get(m), count);
+
+	/*
+	 * Verify dictionary produces correct values
+	 */
+	count2 = 0;
+	for (i=0; i<sizeof(composers)/sizeof(char*); i++) {
+		value = qb_map_get(m, composers[i]);
+		ck_assert_str_eq(value, composers[i]);
+		count2++;
+	}
+	ck_assert_int_eq(qb_map_count_get(m), count2);
+
+	/*
+	 * Delete all dictionary entries
+	 */
+	for (i=0; i<sizeof(composers)/sizeof(char*); i++) {
+		res = qb_map_rm(m, composers[i]);
+		ck_assert_int_eq(res, QB_TRUE);
+	}
+	ck_assert_int_eq(qb_map_count_get(m), 0);
+}
+
 START_TEST(test_skiplist_simple)
 {
 	qb_map_t *m = qb_skiplist_create();
@@ -873,6 +925,31 @@ START_TEST(test_trie_load)
 }
 END_TEST
 
+START_TEST(test_skiplist_accents)
+{
+	qb_map_t *m;
+	m = qb_skiplist_create();
+	test_accents_load(m, __func__);
+}
+END_TEST
+
+START_TEST(test_hashtable_accents)
+{
+	qb_map_t *m;
+	m = qb_hashtable_create(16);
+	test_accents_load(m, __func__);
+}
+END_TEST
+
+START_TEST(test_trie_accents)
+{
+	qb_map_t *m;
+	m = qb_trie_create();
+	test_accents_load(m, __func__);
+}
+END_TEST
+
+
 /*
  * From Honza: https://github.com/asalkeld/libqb/issues/44
  */
@@ -936,6 +1013,10 @@ map_suite(void)
 	add_tcase(s, tc, test_hashtable_load, 30);
 	add_tcase(s, tc, test_trie_load, 30);
 
+	add_tcase(s, tc, test_skiplist_accents);
+	add_tcase(s, tc, test_hashtable_accents);
+	add_tcase(s, tc, test_trie_accents);
+
 	return s;
 }
 

From 09e6c8b582bb8ffc3d05ba5c29ad442b48181c41 Mon Sep 17 00:00:00 2001
From: Christine Caulfield <ccaulfie@redhat.com>
Date: Thu, 5 Mar 2020 15:42:37 +0000
Subject: [PATCH 3/3] trie: Some fixes and extra tests from Honza

---
 lib/trie.c        | 4 ++--
 tests/check_map.c | 4 +++-
 2 files changed, 5 insertions(+), 3 deletions(-)

diff --git a/lib/trie.c b/lib/trie.c
index 5b1f8b019..0ecef6390 100644
--- a/lib/trie.c
+++ b/lib/trie.c
@@ -64,8 +64,8 @@ static void trie_destroy_node(struct trie_node *node);
  * characters are stored in reverse to make accessing the
  * more common case (non-control chars) more space efficient.
  */
-#define TRIE_CHAR2INDEX(ch) abs(126 - ch)
-#define TRIE_INDEX2CHAR(idx) abs(126 - idx)
+#define TRIE_CHAR2INDEX(ch) (127 - (signed char)ch)
+#define TRIE_INDEX2CHAR(idx) (127 - (signed char)idx)
 
 
 static int32_t
diff --git a/tests/check_map.c b/tests/check_map.c
index 578e19bdf..5a98f757c 100644
--- a/tests/check_map.c
+++ b/tests/check_map.c
@@ -50,7 +50,9 @@ const char *composers[] = {
 	"Benjamin Britten",
 	"Josef Haydn",
 	"Claude Debussy",
-	"Charles Ives"
+	"Charles Ives",
+	/* Maybe in an alien language ... but they can cause trie crashes & conflicts */
+	"\x7e\x7f\x80\x81", "\x7e", "\x7e\x7f", "\x7e\x7f\x80",
 };