UTF-8 validation tests and fixes.

lib/utf8_mosq.c

@@ -80,9 +80,13 @@ int mosquitto_validate_utf8(const char *str, int len)
 		}
 
 		/* Check for overlong or out of range encodings */
-		if(codelen == 2 && codepoint < 0x0080){
-			return MOSQ_ERR_MALFORMED_UTF8;
-		}else if(codelen == 3 && codepoint < 0x0800){
+		/* Checking codelen == 2 isn't necessary here, because it is already
+		 * covered above in the C0 and C1 checks.
+		 * if(codelen == 2 && codepoint < 0x0080){
+		 *	 return MOSQ_ERR_MALFORMED_UTF8;
+		 * }else
+		*/
+		if(codelen == 3 && codepoint < 0x0800){
 			return MOSQ_ERR_MALFORMED_UTF8;
 		}else if(codelen == 4 && (codepoint < 0x10000 || codepoint > 0x10FFFF)){
 			return MOSQ_ERR_MALFORMED_UTF8;

test/unit/test.c

@@ -4,6 +4,7 @@
 #include <CUnit/Basic.h>
 
 int init_datatype_read_tests(void);
+int init_datatype_write_tests(void);
 int init_utf8_tests(void);
 
 int main(int argc, char *argv[])

test/unit/utf8.c

@@ -35,6 +35,22 @@ static void TEST_utf8_valid(void)
 }
 
 
+static void TEST_utf8_truncated(void)
+{
+	char buf[4];
+
+	/* As per boundary condition tests, but less one character */
+	buf[0] = 0xC2; buf[1] = 0;
+	utf8_helper(buf, MOSQ_ERR_MALFORMED_UTF8);
+
+	buf[0] = 0xE0; buf[1] = 0xA0; buf[2] = 0;
+	utf8_helper(buf, MOSQ_ERR_MALFORMED_UTF8);
+
+	buf[0] = 0xF0; buf[1] = 0x90; buf[2] = 0x80; buf[3] = 0;
+	utf8_helper(buf, MOSQ_ERR_MALFORMED_UTF8);
+}
+
+
 static void TEST_utf8_boundary_conditions(void)
 {
 	/* 2  Boundary condition test cases */
@@ -285,6 +301,7 @@ int init_utf8_tests(void)
 	if(0
 			|| !CU_add_test(test_suite, "UTF-8 empty", TEST_utf8_empty)
 			|| !CU_add_test(test_suite, "UTF-8 valid", TEST_utf8_valid)
+			|| !CU_add_test(test_suite, "UTF-8 truncated", TEST_utf8_truncated)
 			|| !CU_add_test(test_suite, "UTF-8 boundary conditions", TEST_utf8_boundary_conditions)
 			|| !CU_add_test(test_suite, "UTF-8 malformed sequences", TEST_utf8_malformed_sequences)
 			|| !CU_add_test(test_suite, "UTF-8 overlong encoding", TEST_utf8_overlong_encoding)