public void init() {
try {
objectQueue = new ArrayBlockingQueue<>(SPOOLER_QUEUE_SIZE);
spoolQueueMeter = context.createMeter("spoolQueue");
pathMatcher = new AntPathMatcher(s3ConfigBean.s3Config.delimiter);
} catch (Exception ex) {
throw new RuntimeException(ex);
}
}
private void handleException(Source.Context context, String messageId, Exception ex)
throws StageException {
switch (context.getOnErrorRecord()) {
case DISCARD:
break;
case TO_ERROR:
context.reportError(ParserErrors.PARSER_03, messageId, ex.toString(), ex);
break;
case STOP_PIPELINE:
if (ex instanceof StageException) {
throw (StageException) ex;
} else {
throw new StageException(ParserErrors.PARSER_03, messageId, ex.toString(), ex);
}
default:
throw new IllegalStateException(
Utils.format("Unknown on error value '{}'", context.getOnErrorRecord(), ex));
}
}
public List<Stage.ConfigIssue> init(Source.Context context) {
List<Stage.ConfigIssue> issues = new ArrayList<>();
switch (dataFormat) {
case JSON:
if (dataFormatConfig.jsonMaxObjectLen < 1) {
issues.add(
context.createConfigIssue(
DataFormat.JSON.name(),
"dataFormatConfig.maxJsonObjectLen",
ParserErrors.PARSER_04));
}
break;
case TEXT:
if (dataFormatConfig.textMaxLineLen < 1) {
issues.add(
context.createConfigIssue(
DataFormat.TEXT.name(),
"dataFormatConfig.maxLogLineLength",
ParserErrors.PARSER_04));
}
break;
case DELIMITED:
if (dataFormatConfig.csvMaxObjectLen < 1) {
issues.add(
context.createConfigIssue(
DataFormat.DELIMITED.name(),
"dataFormatConfig.csvMaxObjectLen",
ParserErrors.PARSER_04));
}
break;
case XML:
if (messageConfig != null && messageConfig.produceSingleRecordPerMessage) {
issues.add(
context.createConfigIssue(
parentName,
"messageConfig.produceSingleRecordPerMessage",
ParserErrors.PARSER_06));
}
if (dataFormatConfig.xmlMaxObjectLen < 1) {
issues.add(
context.createConfigIssue(
DataFormat.XML.name(),
"dataFormatConfig.maxXmlObjectLen",
ParserErrors.PARSER_04));
}
if (dataFormatConfig.xmlRecordElement != null
&& !dataFormatConfig.xmlRecordElement.isEmpty()
&& !XMLChar.isValidName(dataFormatConfig.xmlRecordElement)) {
issues.add(
context.createConfigIssue(
DataFormat.XML.name(),
"dataFormatConfig.xmlRecordElement",
ParserErrors.PARSER_02,
dataFormatConfig.xmlRecordElement));
}
break;
case SDC_JSON:
break;
case LOG:
logDataFormatValidator =
new LogDataFormatValidator(
dataFormatConfig.logMode,
dataFormatConfig.logMaxObjectLen,
dataFormatConfig.retainOriginalLine,
dataFormatConfig.customLogFormat,
dataFormatConfig.regex,
dataFormatConfig.grokPatternDefinition,
dataFormatConfig.grokPattern,
dataFormatConfig.enableLog4jCustomLogFormat,
dataFormatConfig.log4jCustomLogFormat,
dataFormatConfig.onParseError,
dataFormatConfig.maxStackTraceLines,
DataFormat.LOG.name(),
getFieldPathToGroupMap(dataFormatConfig.fieldPathsToGroupName));
logDataFormatValidator.validateLogFormatConfig(issues, context);
break;
case AVRO:
if (!dataFormatConfig.schemaInMessage
&& (dataFormatConfig.avroSchema == null || dataFormatConfig.avroSchema.isEmpty())) {
issues.add(
context.createConfigIssue(
DataFormat.AVRO.name(),
"dataFormatConfig.avroSchema",
ParserErrors.PARSER_07,
dataFormatConfig.avroSchema));
}
break;
default:
issues.add(
context.createConfigIssue(
parentName, "dataFormat", ParserErrors.PARSER_05, dataFormat));
}
DataParserFactoryBuilder builder =
new DataParserFactoryBuilder(context, dataFormat.getParserFormat())
.setCharset(Charset.defaultCharset());
if (dataFormatConfig.charset == null) {
messageCharset = StandardCharsets.UTF_8;
} else {
try {
messageCharset = Charset.forName(dataFormatConfig.charset);
} catch (UnsupportedCharsetException ex) {
// setting it to a valid one so the parser factory can be configured and tested for more
// errors
messageCharset = StandardCharsets.UTF_8;
issues.add(
context.createConfigIssue(
parentName, "charset", ParserErrors.PARSER_01, dataFormatConfig.charset));
}
}
builder.setCharset(messageCharset).setRemoveCtrlChars(dataFormatConfig.removeCtrlChars);
switch (dataFormat) {
case TEXT:
builder.setMaxDataLen(dataFormatConfig.textMaxLineLen);
break;
case JSON:
builder.setMode(dataFormatConfig.jsonContent);
builder.setMaxDataLen(dataFormatConfig.jsonMaxObjectLen);
break;
case DELIMITED:
builder
.setMaxDataLen(dataFormatConfig.csvMaxObjectLen)
.setMode(dataFormatConfig.csvFileFormat)
.setMode(dataFormatConfig.csvHeader)
.setMode(dataFormatConfig.csvRecordType)
.setConfig(
DelimitedDataParserFactory.DELIMITER_CONFIG, dataFormatConfig.csvCustomDelimiter)
.setConfig(DelimitedDataParserFactory.ESCAPE_CONFIG, dataFormatConfig.csvCustomEscape)
.setConfig(DelimitedDataParserFactory.QUOTE_CONFIG, dataFormatConfig.csvCustomQuote);
break;
case XML:
builder.setMaxDataLen(dataFormatConfig.xmlMaxObjectLen);
builder.setConfig(
XmlDataParserFactory.RECORD_ELEMENT_KEY, dataFormatConfig.xmlRecordElement);
break;
case SDC_JSON:
builder.setMaxDataLen(-1);
break;
case LOG:
logDataFormatValidator.populateBuilder(builder);
break;
case AVRO:
builder
.setMaxDataLen(Integer.MAX_VALUE)
.setConfig(AvroDataParserFactory.SCHEMA_KEY, dataFormatConfig.avroSchema)
.setConfig(
AvroDataParserFactory.SCHEMA_IN_MESSAGE_KEY, dataFormatConfig.schemaInMessage);
break;
default:
throw new IllegalStateException("Unknown data format: " + dataFormat);
}
parserFactory = builder.build();
return issues;
}
