feat(connector): Add topic to mqtt additional columns

integration_tests/mqtt/create_source.sql

@@ -4,7 +4,7 @@ CREATE TABLE
 CREATE TABLE mqtt_source_table
 (
   id integer,
-  name varchar,
+  name varchar
 )
 WITH (
     connector='mqtt',

proto/plan_common.proto

@@ -264,6 +264,8 @@ message AdditionalCollectionName {}
 
 message AdditionalColumnPayload {}
 
+message AdditionalColumnTopic {}
+
 // this type means we read all headers as a whole
 message AdditionalColumnHeaders {}
 
@@ -281,6 +283,7 @@ message AdditionalColumn {
     AdditionalTableName table_name = 10;
     AdditionalCollectionName collection_name = 11;
     AdditionalColumnPayload payload = 12;
+    AdditionalColumnTopic topic = 13;
   }
 }
 
@@ -294,4 +297,5 @@ enum AdditionalColumnType {
   ADDITIONAL_COLUMN_TYPE_FILENAME = 6;
   ADDITIONAL_COLUMN_TYPE_NORMAL = 7;
   ADDITIONAL_COLUMN_TYPE_PAYLOAD = 8;
+  ADDITIONAL_COLUMN_TYPE_TOPIC = 9;
 }

src/connector/src/parser/additional_columns.rs

@@ -23,14 +23,14 @@ use risingwave_pb::plan_common::{
     AdditionalCollectionName, AdditionalColumn, AdditionalColumnFilename, AdditionalColumnHeader,
     AdditionalColumnHeaders, AdditionalColumnKey, AdditionalColumnOffset,
     AdditionalColumnPartition, AdditionalColumnPayload, AdditionalColumnTimestamp,
-    AdditionalDatabaseName, AdditionalSchemaName, AdditionalTableName,
+    AdditionalColumnTopic, AdditionalDatabaseName, AdditionalSchemaName, AdditionalTableName,
 };
 
 use crate::error::ConnectorResult;
 use crate::source::cdc::MONGODB_CDC_CONNECTOR;
 use crate::source::{
-    AZBLOB_CONNECTOR, GCS_CONNECTOR, KAFKA_CONNECTOR, KINESIS_CONNECTOR, NATS_CONNECTOR,
-    OPENDAL_S3_CONNECTOR, POSIX_FS_CONNECTOR, PULSAR_CONNECTOR,
+    AZBLOB_CONNECTOR, GCS_CONNECTOR, KAFKA_CONNECTOR, KINESIS_CONNECTOR, MQTT_CONNECTOR,
+    NATS_CONNECTOR, OPENDAL_S3_CONNECTOR, POSIX_FS_CONNECTOR, PULSAR_CONNECTOR,
 };
 
 // Hidden additional columns connectors which do not support `include` syntax.
@@ -87,6 +87,10 @@ pub static COMPATIBLE_ADDITIONAL_COLUMNS: LazyLock<HashMap<&'static str, HashSet
                     "collection_name",
                 ]),
             ),
+            (
+                MQTT_CONNECTOR,
+                HashSet::from(["topic", "offset", "partition"]),
+            ),
         ])
     });
 
@@ -266,6 +270,14 @@ pub fn build_additional_column_desc(
                 )),
             },
         ),
+        "topic" => ColumnDesc::named_with_additional_column(
+            column_name,
+            column_id,
+            DataType::Varchar,
+            AdditionalColumn {
+                column_type: Some(AdditionalColumnType::Topic(AdditionalColumnTopic {})),
+            },
+        ),
         _ => unreachable!(),
     };
 

src/connector/src/parser/mod.rs

@@ -494,6 +494,14 @@ impl SourceStreamChunkRowWriter<'_> {
                     // do special logic in `KvEvent::access_field`
                     parse_field(desc)
                 }
+                (_, &Some(AdditionalColumnType::Topic(_))) => {
+                    // topic is used as partition in mqtt connector
+                    return Ok(A::output_for(
+                        self.row_meta
+                            .as_ref()
+                            .map(|ele| ScalarRefImpl::Utf8(ele.split_id)),
+                    ));
+                }
                 (_, _) => {
                     // For normal columns, call the user provided closure.
                     parse_field(desc)

src/connector/src/source/mqtt/enumerator/mod.rs

@@ -17,17 +17,19 @@ use std::sync::atomic::AtomicBool;
 use std::sync::Arc;
 
 use async_trait::async_trait;
+use risingwave_common::bail;
 use rumqttc::v5::{ConnectionError, Event, Incoming};
 use rumqttc::Outgoing;
 use thiserror_ext::AsReport;
 use tokio::sync::RwLock;
 
 use super::source::MqttSplit;
-use super::{MqttError, MqttProperties};
-use crate::error::{ConnectorError, ConnectorResult};
+use super::MqttProperties;
+use crate::error::ConnectorResult;
 use crate::source::{SourceEnumeratorContextRef, SplitEnumerator};
 
 pub struct MqttSplitEnumerator {
+    #[expect(dead_code)]
     topic: String,
     #[expect(dead_code)]
     client: rumqttc::v5::AsyncClient,
@@ -117,10 +119,18 @@ impl SplitEnumerator for MqttSplitEnumerator {
 
     async fn list_splits(&mut self) -> ConnectorResult<Vec<MqttSplit>> {
         if !self.connected.load(std::sync::atomic::Ordering::Relaxed) {
-            return Err(ConnectorError::from(MqttError(format!(
-                "Failed to connect to MQTT broker for topic {}",
-                self.topic
-            ))));
+            let start = std::time::Instant::now();
+            loop {
+                if self.connected.load(std::sync::atomic::Ordering::Relaxed) {
+                    break;
+                }
+
+                if start.elapsed().as_secs() > 10 {
+                    bail!("Failed to connect to mqtt broker");
+                }
+
+                tokio::time::sleep(std::time::Duration::from_millis(500)).await;
+            }
         }
 
         let topics = self.topics.read().await;

src/connector/src/source/mqtt/source/reader.rs

@@ -52,7 +52,8 @@ impl SplitReader for MqttSplitReader {
     ) -> Result<Self> {
         let (client, eventloop) = properties
             .common
-            .build_client(source_ctx.actor_id, source_ctx.fragment_id as u64)?;
+            .build_client(source_ctx.actor_id, source_ctx.fragment_id as u64)
+            .inspect_err(|e| tracing::error!("Failed to build mqtt client: {}", e.as_report()))?;
 
         let qos = properties.common.qos();
 

src/prost/build.rs

@@ -166,6 +166,7 @@ fn main() -> Result<(), Box<dyn std::error::Error>> {
             "plan_common.AdditionalCollectionName",
             "#[derive(Eq, Hash)]",
         )
+        .type_attribute("plan_common.AdditionalColumnTopic", "#[derive(Eq, Hash)]")
         .type_attribute("plan_common.AsOfJoinDesc", "#[derive(Eq, Hash)]")
         .type_attribute("common.ColumnOrder", "#[derive(Eq, Hash)]")
         .type_attribute("common.OrderType", "#[derive(Eq, Hash)]")

src/sqlparser/tests/testdata/create.yaml

@@ -85,6 +85,17 @@
       topic = 'dummy')
     FORMAT plain ENCODE bytes;
   formatted_sql: CREATE TABLE t (raw BYTEA) INCLUDE header AS all_headers INCLUDE header 'foo' AS foo_bytea INCLUDE header 'foo' CHARACTER VARYING AS foo_str INCLUDE header INCLUDE header 'foo' INCLUDE header 'foo' CHARACTER VARYING INCLUDE header 'foo' BYTEA INCLUDE header 'bar' WITH (connector = 'kafka', kafka.brokers = '${RISEDEV_KAFKA_BOOTSTRAP_SERVERS}', topic = 'dummy') FORMAT PLAIN ENCODE BYTES
+- input: |-
+    CREATE TABLE t
+      (raw BYTEA)
+      INCLUDE topic
+    WITH (
+      connector = 'mqtt',
+      url = 'tcp://mqtt-server',
+      topic = 'test',
+      qos = 'at_least_once')
+    FORMAT plain ENCODE bytes;
+  formatted_sql: CREATE TABLE t (raw BYTEA) INCLUDE topic WITH (connector = 'mqtt', url = 'tcp://mqtt-server', topic = 'test', qos = 'at_least_once') FORMAT PLAIN ENCODE BYTES
 - input: CREATE TABLE T (v1 INT, v2 STRUCT<v1 INT, v2 INT>)
   formatted_sql: CREATE TABLE T (v1 INT, v2 STRUCT<v1 INT, v2 INT>)
 - input: CREATE TABLE T (v1 INT, v2 STRUCT<v1 INT, v2 INT, v3 STRUCT<v1 INT, v2 INT>>)