private boolean validateDataGenerator(
Stage.Context context,
DataFormat dataFormat,
String groupName,
List<Stage.ConfigIssue> issues) {
boolean valid = true;
DataGeneratorFactoryBuilder builder =
new DataGeneratorFactoryBuilder(context, dataFormat.getGeneratorFormat());
if (charset == null || charset.trim().isEmpty()) {
charset = CHARSET_UTF8;
}
Charset cSet;
try {
cSet = Charset.forName(charset);
} catch (UnsupportedCharsetException ex) {
// setting it to a valid one so the parser factory can be configured and tested for more
// errors
cSet = StandardCharsets.UTF_8;
issues.add(
context.createConfigIssue(
groupName, "charset", DataFormatErrors.DATA_FORMAT_05, charset));
valid &= false;
}
builder.setCharset(cSet);
switch (dataFormat) {
case SDC_JSON:
break;
case DELIMITED:
builder.setMode(csvFileFormat);
builder.setMode(csvHeader);
builder.setConfig(DelimitedDataGeneratorFactory.REPLACE_NEWLINES_KEY, csvReplaceNewLines);
builder.setConfig(DelimitedDataConstants.DELIMITER_CONFIG, csvCustomDelimiter);
builder.setConfig(DelimitedDataConstants.ESCAPE_CONFIG, csvCustomEscape);
builder.setConfig(DelimitedDataConstants.QUOTE_CONFIG, csvCustomQuote);
break;
case TEXT:
builder.setConfig(TextDataGeneratorFactory.FIELD_PATH_KEY, textFieldPath);
builder.setConfig(TextDataGeneratorFactory.EMPTY_LINE_IF_NULL_KEY, textEmptyLineIfNull);
break;
case JSON:
builder.setMode(jsonMode);
break;
case AVRO:
Schema schema = null;
Map<String, Object> defaultValues = new HashMap<>();
try {
schema =
new Schema.Parser().setValidate(true).setValidateDefaults(true).parse(avroSchema);
} catch (Exception e) {
issues.add(
context.createConfigIssue(
DataFormatGroups.AVRO.name(),
"avroSchema",
DataFormatErrors.DATA_FORMAT_300,
e.toString(),
e));
valid &= false;
}
if (schema != null) {
try {
defaultValues.putAll(
AvroTypeUtil.getDefaultValuesFromSchema(schema, new HashSet<String>()));
} catch (IOException e) {
issues.add(
context.createConfigIssue(
DataFormatGroups.AVRO.name(),
"avroSchema",
DataFormatErrors.DATA_FORMAT_301,
e.toString(),
e));
valid &= false;
}
}
builder.setConfig(AvroDataGeneratorFactory.SCHEMA_KEY, avroSchema);
builder.setConfig(AvroDataGeneratorFactory.INCLUDE_SCHEMA_KEY, includeSchema);
builder.setConfig(AvroDataGeneratorFactory.DEFAULT_VALUES_KEY, defaultValues);
break;
case BINARY:
builder.setConfig(BinaryDataGeneratorFactory.FIELD_PATH_KEY, binaryFieldPath);
break;
case PROTOBUF:
builder
.setConfig(ProtobufConstants.PROTO_DESCRIPTOR_FILE_KEY, protoDescriptorFile)
.setConfig(ProtobufConstants.MESSAGE_TYPE_KEY, messageType);
break;
default:
// no action needed
break;
}
if (valid) {
try {
dataGeneratorFactory = builder.build();
} catch (Exception ex) {
issues.add(
context.createConfigIssue(
null, null, DataFormatErrors.DATA_FORMAT_201, ex.toString(), ex));
valid &= false;
}
}
return valid;
}
