Write names should not underflow size_t

If the read_name is invalid, we shouldn't try to set a write name.

Author: kddnewton

src/yarp.c

@@ -1508,12 +1508,17 @@ yp_call_node_variable_call_p(yp_call_node_t *node) {
 // Initialize the read name by reading the write name and chopping off the '='.
 static void
 yp_call_write_read_name_init(yp_string_t *read_name, yp_string_t *write_name) {
-    size_t length = write_name->length - 1;
+    if (write_name->length >= 1) {
+        size_t length = write_name->length - 1;
 
-    void *memory = malloc(length);
-    memcpy(memory, write_name->source, length);
+        void *memory = malloc(length);
+        memcpy(memory, write_name->source, length);
 
-    yp_string_owned_init(read_name, (uint8_t *) memory, length);
+        yp_string_owned_init(read_name, (uint8_t *) memory, length);
+    } else {
+        // We can get here if the message was missing because of a syntax error.
+        yp_string_constant_init(read_name, "", 0);
+    }
 }
 
 // Allocate and initialize a new CallAndWriteNode node.

test/yarp/errors_test.rb

@@ -1137,6 +1137,11 @@ def test_duplicated_parameter_names
       assert_errors expected, "def foo(a = 1,b,*c);end", [["Unexpected parameter `*`", 16..17]]
     end
 
+    def test_invalid_message_name
+      result = YARP.parse("+.@foo,+=foo")
+      assert_equal "", result.value.statements.body.first.write_name
+    end
+
     def test_invalid_operator_write_fcall
       source = "foo! += 1"
       assert_errors expression(source), source, [