@Override
public void commit(String offset) throws StageException {
consumer.commit(offset);
}
@Override
public boolean inErrorState() {
return producer.inErrorState() || consumer.inErrorState();
}
@Override
public void errorNotification(Throwable throwable) {
consumer.error(throwable);
}
@Override
public String produce(String lastSourceOffset, int maxBatchSize, BatchMaker batchMaker)
throws StageException {
OffsetAndResult<Map.Entry> offsetAndResult;
if (getContext().isPreview()) {
// we support text and csv today
List<Map.Entry> records = new ArrayList<>();
int count = 0;
Iterator<String> keys = previewBuffer.keySet().iterator();
while (count < maxBatchSize && count < previewBuffer.size() && keys.hasNext()) {
String key = keys.next();
String[] keyParts = key.split("::");
if (count == 0
&& DataFormat.DELIMITED == dataFormat
&& CsvHeader.NO_HEADER != csvHeader
&& keyParts.length > 1
&& keyParts[1].equals("0")) {
// add header
if (CsvHeader.WITH_HEADER == csvHeader) {
records.add(new Pair(previewBuffer.get(key), null));
} else if (CsvHeader.IGNORE_HEADER == csvHeader) {
// this record will be ignored - don't increment the count
}
} else {
records.add(new Pair(key, previewBuffer.get(key)));
count++;
}
}
offsetAndResult = new OffsetAndResult<>(recordsProduced, records);
} else {
offsetAndResult = consumer.take();
}
if (offsetAndResult == null) {
LOG.info("Received null batch, returning null");
return null;
}
String messageId = null;
int count = 0;
String header = null;
for (Map.Entry message : offsetAndResult.getResult()) {
count++;
messageId = String.valueOf(message.getKey());
List<Record> listRecords = null;
if (dataFormat == DataFormat.TEXT) {
listRecords = processMessage(messageId, message.getValue());
} else if (dataFormat == DataFormat.DELIMITED) {
switch (csvHeader) {
case IGNORE_HEADER:
// ignore header by skipping this header string
// [1] - startOffset - [2] - contextKey
String[] offsetContextSplit = messageId.split("::");
if (offsetContextSplit.length > 1 && offsetContextSplit[1].equals("0")) {
break;
}
case NO_HEADER:
listRecords = processMessage(messageId, message.getValue());
break;
case WITH_HEADER:
if (header == null) {
header = messageId;
LOG.debug("Header is: {}", header);
Utils.checkState(
message.getValue() == null,
Utils.formatL(
"Message value for header record should be null, was: '{}'",
message.getValue()));
} else {
listRecords = processMessage(messageId, header + "\n" + message.getValue());
}
break;
default:
String msg = Utils.format("Unrecognized header: '{}'", csvHeader);
LOG.warn(msg);
throw new IllegalStateException(msg);
}
} else if (dataFormat == DataFormat.AVRO) {
listRecords = processMessage(messageId, message.getValue());
} else {
throw new IllegalStateException(Utils.format("Unrecognized data format: '{}'", dataFormat));
}
if (listRecords != null) {
for (Record record : listRecords) {
batchMaker.addRecord(record);
}
}
}
if (count == 0) {
LOG.info("Received no records, returning null");
return null;
}
return Utils.checkNotNull(messageId, "Log error, message ID cannot be null at this point.");
}
