hailin0 commented on code in PR #3409:
URL:
https://github.com/apache/incubator-seatunnel/pull/3409#discussion_r1045237022
##########
seatunnel-core/seatunnel-spark-starter/src/main/java/org/apache/seatunnel/core/starter/spark/execution/TransformExecuteProcessor.java:
##########
@@ -70,13 +82,39 @@ public List<Dataset<Row>> execute(List<Dataset<Row>>
upstreamDataStreams) throws
Dataset<Row> input = upstreamDataStreams.get(0);
List<Dataset<Row>> result = new ArrayList<>();
for (int i = 0; i < plugins.size(); i++) {
- BaseSparkTransform transform = plugins.get(i);
- Config pluginConfig = pluginConfigs.get(i);
- Dataset<Row> stream = fromSourceTable(pluginConfig,
sparkEnvironment).orElse(input);
- input = transform.process(stream, sparkEnvironment);
- registerInputTempView(pluginConfig, input);
- result.add(input);
+ try {
+ SeaTunnelTransform<SeaTunnelRow> transform = plugins.get(i);
+ Config pluginConfig = pluginConfigs.get(i);
+ Dataset<Row> stream = fromSourceTable(pluginConfig,
sparkEnvironment).orElse(input);
+ input = sparkTransform(transform, stream);
+ registerInputTempView(pluginConfig, input);
+ result.add(input);
+ } catch (Exception e) {
+ throw new TaskExecuteException(
+ String.format("SeaTunnel transform task: %s execute
error", plugins.get(i).getPluginName()), e);
+ }
}
return result;
}
+
+ private Dataset<Row> sparkTransform(SeaTunnelTransform transform,
Dataset<Row> stream) throws IOException {
+ SeaTunnelDataType<?> seaTunnelDataType =
TypeConverterUtils.convert(stream.schema());
+ transform.setTypeInfo(seaTunnelDataType);
+ StructType structType = (StructType)
TypeConverterUtils.convert(transform.getProducedType());
+ List<Row> outputRows = new ArrayList<>();
+ Iterator<Row> rowIterator = stream.toLocalIterator();
+ SeaTunnelRow seaTunnelRow;
+ while (rowIterator.hasNext()) {
+ Row row = rowIterator.next();
+ seaTunnelRow = new
InternalRowConverter(seaTunnelDataType).reconvert(InternalRow.apply(row.toSeq()));
+ seaTunnelRow = (SeaTunnelRow) transform.map(seaTunnelRow);
+ InternalRow internalRow = new
InternalRowConverter(transform.getProducedType()).convert(seaTunnelRow);
+ MutableValue[] mutableValues = ((SpecificInternalRow)
internalRow).values();
+ outputRows.add(new GenericRowWithSchema(
+
Arrays.stream(mutableValues).map(MutableValue::boxed).toArray(),
+ structType));
+ }
+ return sparkEnvironment.getSparkSession().createDataFrame(outputRows,
structType);
Review Comment:
maybe should use `Dataset<Row> stream = stream.map(...)` or `Dataset<Row>
stream = stream.transform(...)`?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
