Merge pull request confluentinc#131 from confluentinc/errors_129

edenhill · web-flow · commit ece3df4772f7 · 2017-02-27T08:23:51.000+01:00
Proper error string handling in Producer, issue confluentinc#129
diff --git a/confluent_kafka/src/Consumer.c b/confluent_kafka/src/Consumer.c
@@ -463,7 +463,7 @@ static PyObject *Consumer_poll (Handle *self, PyObject *args,
         if (!rkm)
                 Py_RETURN_NONE;
 
-        msgobj = Message_new0(rkm);
+        msgobj = Message_new0(self, rkm);
         rd_kafka_message_destroy(rkm);
 
         return msgobj;
@@ -770,6 +770,8 @@ static int Consumer_init (PyObject *selfobj, PyObject *args, PyObject *kwargs) {
                 return -1;
         }
 
+        self->type = RD_KAFKA_CONSUMER;
+
         if (!(conf = common_conf_setup(RD_KAFKA_CONSUMER, self,
                                        args, kwargs)))
                 return -1; /* Exception raised by ..conf_setup() */
diff --git a/confluent_kafka/src/Producer.c b/confluent_kafka/src/Producer.c
@@ -152,7 +152,7 @@ static void dr_msg_cb (rd_kafka_t *rk, const rd_kafka_message_t *rkm,
 		goto done;
 	}
 
-	msgobj = Message_new0(rkm);
+	msgobj = Message_new0(self, rkm);
 	
 	args = Py_BuildValue("(OO)",
 			     Message_error((Message *)msgobj, NULL),
@@ -527,6 +527,8 @@ static int Producer_init (PyObject *selfobj, PyObject *args, PyObject *kwargs) {
                 return -1;
         }
 
+        self->type = RD_KAFKA_PRODUCER;
+
         if (!(conf = common_conf_setup(RD_KAFKA_PRODUCER, self,
                                        args, kwargs)))
                 return -1;
diff --git a/confluent_kafka/src/confluent_kafka.c b/confluent_kafka/src/confluent_kafka.c
@@ -258,7 +258,10 @@ PyObject *KafkaError_new0 (rd_kafka_resp_err_t err, const char *fmt, ...) {
  PyObject *KafkaError_new_or_None (rd_kafka_resp_err_t err, const char *str) {
 	if (!err)
 		Py_RETURN_NONE;
-	return KafkaError_new0(err, "%s", str);
+        if (str)
+                return KafkaError_new0(err, "%s", str);
+        else
+                return KafkaError_new0(err, NULL);
 }
 
 
@@ -529,17 +532,19 @@ PyTypeObject MessageType = {
 /**
  * @brief Internal factory to create Message object from message_t
  */
-PyObject *Message_new0 (const rd_kafka_message_t *rkm) {
+PyObject *Message_new0 (const Handle *handle, const rd_kafka_message_t *rkm) {
 	Message *self;
 
 	self = (Message *)MessageType.tp_alloc(&MessageType, 0);
 	if (!self)
 		return NULL;
 
-	self->error = KafkaError_new_or_None(rkm->err,
-					     rkm->err ?
-					     rd_kafka_message_errstr(rkm) :
-					     NULL);
+        /* Only use message error string on Consumer, for Producers
+         * it will contain the original message payload. */
+        self->error = KafkaError_new_or_None(
+                rkm->err,
+                (rkm->err && handle->type != RD_KAFKA_PRODUCER) ?
+                rd_kafka_message_errstr(rkm) : NULL);
 
 	if (rkm->rkt)
 		self->topic = cfl_PyUnistr(
diff --git a/confluent_kafka/src/confluent_kafka.h b/confluent_kafka/src/confluent_kafka.h
@@ -125,6 +125,7 @@ typedef struct {
 	PyObject *stats_cb;
         int initiated;
         int tlskey;  /* Thread-Local-Storage key */
+        rd_kafka_type_t type; /* Producer or consumer */
 
 	union {
 		/**
@@ -255,7 +256,7 @@ typedef struct {
 
 extern PyTypeObject MessageType;
 
-PyObject *Message_new0 (const rd_kafka_message_t *rkm);
+PyObject *Message_new0 (const Handle *handle, const rd_kafka_message_t *rkm);
 PyObject *Message_error (Message *self, PyObject *ignore);
 
 
diff --git a/tests/test_Producer.py b/tests/test_Producer.py
@@ -74,3 +74,27 @@ def produce_hi (self):
     sp = SubProducer({'log.thread.name': True}, 'mytopic')
     sp.produce('someother', value='not hello')
     sp.produce_hi()
+
+
+def test_dr_msg_errstr():
+    """
+    Test that the error string for failed messages works (issue #129).
+    The underlying problem is that librdkafka reuses the message payload
+    for error value on Consumer messages, but on Producer messages the
+    payload is the original payload and no rich error string exists.
+    """
+    p = Producer({"default.topic.config":{"message.timeout.ms":10}})
+
+    def handle_dr (err, msg):
+        # Neither message payloads must not affect the error string.
+        assert err is not None
+        assert err.code() == KafkaError._MSG_TIMED_OUT
+        assert "Message timed out" in err.str()
+
+    # Unicode safe string
+    p.produce('mytopic', "This is the message payload", on_delivery=handle_dr)
+
+    # Invalid unicode sequence
+    p.produce('mytopic', "\xc2\xc2", on_delivery=handle_dr)
+
+    p.flush()

Original file line number	Diff line number	Diff line change
`@@ -152,7 +152,7 @@ static void dr_msg_cb (rd_kafka_t rk, const rd_kafka_message_t rkm,`
`152`	`152`	`goto done;`
`153`	`153`	`}`
`154`	`154`
`155`		`- msgobj = Message_new0(rkm);`
	`155`	`+ msgobj = Message_new0(self, rkm);`
`156`	`156`
`157`	`157`	`args = Py_BuildValue("(OO)",`
`158`	`158`	`Message_error((Message *)msgobj, NULL),`
`@@ -527,6 +527,8 @@ static int Producer_init (PyObject selfobj, PyObject args, PyObject *kwargs) {`
`527`	`527`	`return -1;`
`528`	`528`	`}`
`529`	`529`
	`530`	`+ self->type = RD_KAFKA_PRODUCER;`
	`531`	`+`
`530`	`532`	`if (!(conf = common_conf_setup(RD_KAFKA_PRODUCER, self,`
`531`	`533`	`args, kwargs)))`
`532`	`534`	`return -1;`