@Test
public void testEmptyBatch() throws Exception {
List<JdbcFieldColumnMapping> columnMappings =
ImmutableList.of(new JdbcFieldColumnMapping("P_ID", "[3]"));
JdbcLookupDProcessor processor = new JdbcLookupDProcessor();
processor.hikariConfigBean = createConfigBean(h2ConnectionString, username, password);
ProcessorRunner processorRunner =
new ProcessorRunner.Builder(JdbcLookupDProcessor.class, processor)
.addConfiguration("query", listQuery)
.addConfiguration("columnMappings", columnMappings)
.addConfiguration("maxClobSize", 1000)
.addConfiguration("maxBlobSize", 1000)
.addOutputLane("lane")
.build();
List<Record> emptyBatch = ImmutableList.of();
processorRunner.runInit();
try {
StageRunner.Output output = processorRunner.runProcess(emptyBatch);
Assert.assertEquals(0, output.getRecords().get("lane").size());
} finally {
processorRunner.runDestroy();
}
}
@Test
public void testProduceTarGzipAvroFile() throws Exception {
AmazonS3Source source = createTarGzipAvroSource();
SourceRunner runner =
new SourceRunner.Builder(AmazonS3DSource.class, source).addOutputLane("lane").build();
runner.runInit();
try {
List<Record> allRecords = new ArrayList<>();
String offset = null;
for (int i = 0; i < 50; i++) {
BatchMaker batchMaker = SourceRunner.createTestBatchMaker("lane");
offset = source.produce(offset, 1000, batchMaker);
Assert.assertNotNull(offset);
StageRunner.Output output = SourceRunner.getOutput(batchMaker);
List<Record> records = output.getRecords().get("lane");
allRecords.addAll(records);
}
Assert.assertEquals(48000, allRecords.size());
Assert.assertTrue(
offset.contains("NorthAmerica/testAvro2.tar.gz::-1::c17d97fdd6f2c6902efe059753cf41b6::"));
} finally {
runner.runDestroy();
}
}
@Test
public void testCategoryRegexInNonComplexType() throws StageException {
FieldRenamerConfig renameConfig = new FieldRenamerConfig();
// Any field containing a non-word character should be in single quotes
renameConfig.fromFieldExpression = "/'(.*)[#&@|](.*)'";
renameConfig.toFieldExpression = "/$1_$2";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.CONTINUE;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.APPEND_NUMBERS;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig), errorHandler);
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.setOnRecordError(OnRecordError.TO_ERROR)
.addOutputLane("a")
.build();
runner.runInit();
try {
Map<String, Field> map = new LinkedHashMap<>();
map.put("a#b", Field.create(Field.Type.STRING, "foo1"));
map.put("a_b", Field.create(Field.Type.STRING, "foo2"));
map.put("a&b", Field.create(Field.Type.STRING, "foo3"));
map.put("a|b", Field.create(Field.Type.STRING, "foo4"));
map.put("a@b", Field.create(Field.Type.STRING, "foo5"));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(Field.Type.MAP, map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record r = output.getRecords().get("a").get(0);
Assert.assertFalse(r.has("/'a#b'"));
Assert.assertFalse(r.has("/'a&b'"));
Assert.assertFalse(r.has("/'a|b'"));
Assert.assertFalse(r.has("/'a&b'"));
Assert.assertTrue(r.has("/a_b"));
Assert.assertEquals("foo2", r.get("/a_b").getValueAsString());
Assert.assertTrue(r.has("/a_b1"));
Assert.assertEquals("foo1", r.get("/a_b1").getValueAsString());
Assert.assertTrue(r.has("/a_b2"));
Assert.assertEquals("foo3", r.get("/a_b2").getValueAsString());
Assert.assertTrue(r.has("/a_b3"));
Assert.assertEquals("foo4", r.get("/a_b3").getValueAsString());
Assert.assertTrue(r.has("/a_b4"));
Assert.assertEquals("foo5", r.get("/a_b4").getValueAsString());
} finally {
runner.runDestroy();
}
}
@Test
public void testProduceFullFile() throws Exception {
AmazonS3Source source = createSource();
SourceRunner runner =
new SourceRunner.Builder(AmazonS3DSource.class, source).addOutputLane("lane").build();
runner.runInit();
try {
int initialCount = getObjectCount(s3client, BUCKET_NAME);
int postProcessInitialCount = getObjectCount(s3client, POSTPROCESS_BUCKET);
List<Record> allRecords = new ArrayList<>();
String offset = null;
for (int i = 0; i < 10; i++) {
BatchMaker batchMaker = SourceRunner.createTestBatchMaker("lane");
offset = source.produce(offset, 60000, batchMaker);
Assert.assertNotNull(offset);
StageRunner.Output output = SourceRunner.getOutput(batchMaker);
List<Record> records = output.getRecords().get("lane");
allRecords.addAll(records);
}
Assert.assertEquals(6, allRecords.size());
Assert.assertEquals(initialCount - 6, getObjectCount(s3client, BUCKET_NAME));
Assert.assertEquals(
postProcessInitialCount + 6, getObjectCount(s3client, POSTPROCESS_BUCKET));
} finally {
runner.runDestroy();
}
}
@Test
public void testNestedKeep3() throws StageException {
/*
* try to retain the second element from the root array list.
* Make sure that this turns out to be the first element in the resulting record
*/
Record record = createNestedRecord();
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration(
"fields", ImmutableList.of("/USA[1]/SantaMonica/cole/streets[0][1]/name"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertTrue(resultRecord.has("/USA[0]/SantaMonica/cole/streets[0][0]/name"));
Assert.assertEquals(
"f", resultRecord.get("/USA[0]/SantaMonica/cole/streets[0][0]/name").getValueAsString());
Assert.assertFalse(resultRecord.has("/USA[0]/SantaMonica/cole/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco"));
Assert.assertFalse(resultRecord.has("/USA[1]"));
} finally {
runner.runDestroy();
}
}
@Test
public void testNestedKeep5() throws StageException {
Record record = createNestedRecord();
/*
* keep all entries of a list by specifying path just upto the list
*/
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration("fields", ImmutableList.of("/USA[0]/SanFrancisco/folsom/streets[0]"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0]"));
// Its a list and it is empty. use wild card [*] to preserve the contents of the list
Assert.assertEquals(
2, resultRecord.get("/USA[0]/SanFrancisco/folsom/streets[0]").getValueAsList().size());
} finally {
runner.runDestroy();
}
}
@Test
public void testNestedKeep6() throws StageException {
Record record = createNestedRecord();
/*
* keep all entries of a map by specifying path just upto the map
*/
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration("fields", ImmutableList.of("/USA[0]/SanFrancisco/noe"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe"));
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[0][0]/name").getValueAsString(), "a");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[0][1]/name").getValueAsString(), "b");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[1][0]/name").getValueAsString(), "c");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[1][1]/name").getValueAsString(), "d");
} finally {
runner.runDestroy();
}
}
@Test
public void testWildCardRemove1() throws StageException {
Record record = createNestedRecord();
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration(
"fields", ImmutableList.of("/USA[*]/SanFrancisco/*/streets[*][*]/name"))
.addConfiguration("filterOperation", FilterOperation.REMOVE)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[1][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[1][1]/name"));
Assert.assertEquals(
resultRecord.get("/USA[1]/SantaMonica/cole/streets[0][0]/name").getValueAsString(), "e");
Assert.assertEquals(
resultRecord.get("/USA[1]/SantaMonica/cole/streets[0][1]/name").getValueAsString(), "f");
} finally {
runner.runDestroy();
}
}
@Test
public void testKeepNonExistingFiled() throws StageException {
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration("fields", ImmutableList.of("/city"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
Map<String, Field> map = new LinkedHashMap<>();
map.put("name", Field.create("a"));
map.put("age", Field.create("b"));
map.put("streetAddress", Field.create("c"));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Field field = output.getRecords().get("a").get(0).get();
Assert.assertTrue(field.getValue() instanceof Map);
Map<String, Field> result = field.getValueAsMap();
Assert.assertTrue(result.size() == 0);
} finally {
runner.runDestroy();
}
}
@Test
public void testProduceTarGzipTextFile() throws Exception {
AmazonS3Source source = createTarGzipSource();
SourceRunner runner =
new SourceRunner.Builder(AmazonS3DSource.class, source).addOutputLane("lane").build();
runner.runInit();
try {
List<Record> allRecords = new ArrayList<>();
String offset = null;
for (int i = 0; i < 50; i++) {
BatchMaker batchMaker = SourceRunner.createTestBatchMaker("lane");
offset = source.produce(offset, 1000, batchMaker);
Assert.assertNotNull(offset);
StageRunner.Output output = SourceRunner.getOutput(batchMaker);
List<Record> records = output.getRecords().get("lane");
allRecords.addAll(records);
}
Assert.assertEquals(37044, allRecords.size());
Assert.assertTrue(
offset.contains(
"NorthAmerica/logArchive2.tar.gz::-1::9c91073f2c2b51ed80c0a33da1238214::"));
} finally {
runner.runDestroy();
}
}
@Test
public void testMultipleRegexMatchingSameField() throws StageException {
FieldRenamerConfig renameConfig1 = new FieldRenamerConfig();
renameConfig1.fromFieldExpression = "/sql(.*)";
renameConfig1.toFieldExpression = "/sqlRename$1";
FieldRenamerConfig renamerConfig2 = new FieldRenamerConfig();
renamerConfig2.fromFieldExpression = "/s(.*)";
renamerConfig2.toFieldExpression = "/sRename$1";
Map<String, Field> map = new LinkedHashMap<>();
map.put("sqlField", Field.create(Field.Type.STRING, "foo"));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(Field.Type.MAP, map));
try {
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.CONTINUE;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.TO_ERROR;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig1, renamerConfig2), errorHandler);
// Test non-existent source with existing target field
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.setOnRecordError(OnRecordError.STOP_PIPELINE)
.addOutputLane("a")
.build();
runner.runInit();
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.fail("Should throw error if multiple regex match the same field");
} catch (OnRecordErrorException e) {
Assert.assertEquals(Errors.FIELD_RENAMER_03, e.getErrorCode());
}
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.CONTINUE;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.CONTINUE;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.TO_ERROR;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig1, renamerConfig2), errorHandler);
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.setOnRecordError(OnRecordError.TO_ERROR)
.addOutputLane("a")
.build();
runner.runInit();
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record r = output.getRecords().get("a").get(0);
Assert.assertTrue(r.has("/sqlField"));
}
@Test
public void testRenameMultipleListElementsWithConstIdxExpr() throws StageException {
FieldRenamerConfig renameConfig1 = new FieldRenamerConfig();
renameConfig1.fromFieldExpression = "/listOfInts[0]";
renameConfig1.toFieldExpression = "/nonExisting0";
FieldRenamerConfig renameConfig2 = new FieldRenamerConfig();
renameConfig2.fromFieldExpression = "/listOfInts[1]";
renameConfig2.toFieldExpression = "/nonExisting1";
FieldRenamerConfig renameConfig3 = new FieldRenamerConfig();
renameConfig3.fromFieldExpression = "/listOfInts[2]";
renameConfig3.toFieldExpression = "/nonExisting2";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.TO_ERROR;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.REPLACE;
// Reverse order in configuration so as to preserve array indices
FieldRenamerProcessor processor =
new FieldRenamerProcessor(
ImmutableList.of(renameConfig3, renameConfig2, renameConfig1), errorHandler);
// Test non-existent source with existing target field
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.addOutputLane("a")
.build();
runner.runInit();
try {
Map<String, Field> map = new LinkedHashMap<>();
map.put(
"listOfInts",
Field.create(
ImmutableList.of(
Field.create(Field.Type.INTEGER, 1),
Field.create(Field.Type.INTEGER, 2),
Field.create(Field.Type.INTEGER, 3))));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Map<String, Field> result = output.getRecords().get("a").get(0).get().getValueAsMap();
Assert.assertTrue(result.containsKey("listOfInts"));
Assert.assertTrue(result.get("listOfInts").getValueAsList().isEmpty());
Assert.assertTrue(result.containsKey("nonExisting0"));
Assert.assertTrue(result.containsKey("nonExisting1"));
Assert.assertTrue(result.containsKey("nonExisting2"));
Assert.assertEquals(1, result.get("nonExisting0").getValueAsInteger());
Assert.assertEquals(2, result.get("nonExisting1").getValueAsInteger());
Assert.assertEquals(3, result.get("nonExisting2").getValueAsInteger());
} finally {
runner.runDestroy();
}
}
@Test
public void testLookup() throws Exception {
String ip = "128.101.101.101";
List<GeolocationFieldConfig> configs = new ArrayList<>();
GeolocationFieldConfig config;
config = new GeolocationFieldConfig();
config.inputFieldName = "/ipAsInt";
config.outputFieldName = "/intIpCountry";
config.targetType = GeolocationField.COUNTRY_NAME;
configs.add(config);
config = new GeolocationFieldConfig();
config.inputFieldName = "/ipAsIntString";
config.outputFieldName = "/intStringIpCountry";
config.targetType = GeolocationField.COUNTRY_NAME;
configs.add(config);
config = new GeolocationFieldConfig();
config.inputFieldName = "/ipAsString";
config.outputFieldName = "/stringIpCountry";
config.targetType = GeolocationField.COUNTRY_NAME;
configs.add(config);
ProcessorRunner runner =
new ProcessorRunner.Builder(GeolocationDProcessor.class)
.addConfiguration("fieldTypeConverterConfigs", configs)
.addConfiguration("geoIP2DBFile", databaseFile.getAbsolutePath())
.addOutputLane("a")
.build();
runner.runInit();
try {
Map<String, Field> map = new LinkedHashMap<>();
map.put("ipAsInt", Field.create(GeolocationProcessor.ipAsStringToInt(ip)));
map.put(
"ipAsIntString", Field.create(String.valueOf(GeolocationProcessor.ipAsStringToInt(ip))));
map.put("ipAsString", Field.create(ip));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(0, runner.getErrorRecords().size());
Assert.assertEquals(1, output.getRecords().get("a").size());
Field field = output.getRecords().get("a").get(0).get();
Assert.assertTrue(field.getValue() instanceof Map);
Map<String, Field> result = field.getValueAsMap();
Assert.assertEquals(String.valueOf(result), 6, result.size());
Assert.assertEquals(
"United States",
Utils.checkNotNull(result.get("intStringIpCountry"), "intStringIpCountry").getValue());
Assert.assertEquals(
"United States",
Utils.checkNotNull(result.get("intIpCountry"), "intIpCountry").getValue());
Assert.assertEquals(
"United States",
Utils.checkNotNull(result.get("stringIpCountry"), "stringIpCountry").getValue());
} finally {
runner.runDestroy();
}
}
@Test(timeout = 30000)
public void testProduce() throws Exception {
SourceRunner sourceRunner =
new SourceRunner.Builder(ClusterHdfsDSource.class)
.addOutputLane("lane")
.setExecutionMode(ExecutionMode.CLUSTER_BATCH)
.addConfiguration("hdfsUri", miniDFS.getURI().toString())
.addConfiguration("hdfsDirLocations", Arrays.asList(dir.toUri().getPath()))
.addConfiguration("recursive", false)
.addConfiguration("hdfsConfigs", new HashMap<String, String>())
.addConfiguration("dataFormat", DataFormat.TEXT)
.addConfiguration("textMaxLineLen", 1024)
.addConfiguration("produceSingleRecordPerMessage", false)
.addConfiguration("regex", null)
.addConfiguration("grokPatternDefinition", null)
.addConfiguration("enableLog4jCustomLogFormat", false)
.addConfiguration("customLogFormat", null)
.addConfiguration("fieldPathsToGroupName", null)
.addConfiguration("log4jCustomLogFormat", null)
.addConfiguration("grokPattern", null)
.addConfiguration("hdfsKerberos", false)
.addConfiguration("hdfsConfDir", hadoopConfDir)
.setResourcesDir(resourcesDir)
.build();
sourceRunner.runInit();
List<Map.Entry> list = new ArrayList<>();
list.add(new Pair(new LongWritable(1), new Text("aaa")));
list.add(new Pair(new LongWritable(2), new Text("bbb")));
list.add(new Pair(new LongWritable(3), new Text("ccc")));
Thread th = createThreadForAddingBatch(sourceRunner, list);
try {
StageRunner.Output output = sourceRunner.runProduce(null, 5);
String newOffset = output.getNewOffset();
Assert.assertEquals("3", newOffset);
List<Record> records = output.getRecords().get("lane");
Assert.assertEquals(3, records.size());
for (int i = 0; i < records.size(); i++) {
Assert.assertNotNull(records.get(i).get("/text"));
LOG.info("Header " + records.get(i).getHeader().getSourceId());
Assert.assertTrue(!records.get(i).get("/text").getValueAsString().isEmpty());
Assert.assertEquals(
list.get(i).getValue().toString(), records.get(i).get("/text").getValueAsString());
}
if (sourceRunner != null) {
sourceRunner.runDestroy();
}
} finally {
th.interrupt();
}
}
@Test
public void testDifferentMatchesRegex() throws StageException {
FieldRenamerConfig renameConfig = new FieldRenamerConfig();
// Any field containing a non-word character should be in single quotes
renameConfig.fromFieldExpression = "/'(.*)(#)(.*)'";
renameConfig.toFieldExpression = "/$1hash$3";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.TO_ERROR;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.TO_ERROR;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig), errorHandler);
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.setOnRecordError(OnRecordError.STOP_PIPELINE)
.addOutputLane("a")
.build();
runner.runInit();
Map<String, Field> map = new LinkedHashMap<>();
map.put("#abcd", Field.create("hashabcd"));
Record record1 = RecordCreator.create("s", "s:1");
record1.set(Field.create(Field.Type.MAP, map));
map = new LinkedHashMap<>();
map.put("ab#cd", Field.create("abhashcd"));
Record record2 = RecordCreator.create("s", "s:2");
record2.set(Field.create(Field.Type.MAP, map));
map = new LinkedHashMap<>();
map.put("abcd#", Field.create("abcdhash"));
Record record3 = RecordCreator.create("s", "s:3");
record3.set(Field.create(Field.Type.MAP, map));
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record1, record2, record3));
Assert.assertEquals(3, output.getRecords().get("a").size());
for (Record record : output.getRecords().get("a")) {
Map<String, Field> fieldMap = record.get().getValueAsMap();
for (Map.Entry<String, Field> fieldEntry : fieldMap.entrySet()) {
Assert.assertEquals(fieldEntry.getKey(), fieldEntry.getValue().getValueAsString());
}
}
} finally {
runner.runDestroy();
}
}
@Test
public void testRegexInComplexListType() throws StageException {
FieldRenamerConfig renameConfig = new FieldRenamerConfig();
// Any field containing a non-word character should be in single quotes
renameConfig.fromFieldExpression = "/(*)[(*)]/'SQL(#)(.*)'";
renameConfig.toFieldExpression = "/$1[$2]/SQL$4";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.TO_ERROR;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.TO_ERROR;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig), errorHandler);
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.setOnRecordError(OnRecordError.TO_ERROR)
.addOutputLane("a")
.build();
runner.runInit();
try {
Map<String, Field> innerMap1 = new LinkedHashMap<>();
innerMap1.put("SQL#1", Field.create(Field.Type.STRING, "foo1"));
Map<String, Field> innerMap2 = new LinkedHashMap<>();
innerMap2.put("SQL#2", Field.create(Field.Type.STRING, "foo2"));
List<Field> list = new LinkedList<>();
list.add(Field.create(innerMap1));
list.add(Field.create(innerMap2));
Map<String, Field> map = new HashMap<>();
map.put("list", Field.create(list));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record r = output.getRecords().get("a").get(0);
Assert.assertFalse(r.getEscapedFieldPaths().contains("/list[0]/'SQL#1'"));
Assert.assertFalse(r.getEscapedFieldPaths().contains("/list[1]/'SQL#2'"));
Assert.assertTrue(r.getEscapedFieldPaths().contains("/list[0]/SQL1"));
Assert.assertTrue(r.getEscapedFieldPaths().contains("/list[1]/SQL2"));
} finally {
runner.runDestroy();
}
}
@Test
public void testTargetFieldExistsAppendNumbers() throws StageException {
FieldRenamerConfig renameConfig = new FieldRenamerConfig();
// Any field containing a non-word character should be in single quotes
renameConfig.fromFieldExpression = "/field";
renameConfig.toFieldExpression = "/col";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.CONTINUE;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.APPEND_NUMBERS;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig), errorHandler);
// Test non-existent source with existing target field
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.addOutputLane("a")
.build();
runner.runInit();
try {
Map<String, Field> map = new LinkedHashMap<>();
map.put("field", Field.create(Field.Type.STRING, "field"));
map.put("col", Field.create(Field.Type.STRING, "col"));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(Field.Type.MAP, map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record r = output.getRecords().get("a").get(0);
Assert.assertFalse(r.has("/field"));
Assert.assertTrue(r.has("/col"));
Assert.assertTrue(r.has("/col1"));
Assert.assertEquals("col", r.get("/col").getValueAsString());
Assert.assertEquals("field", r.get("/col1").getValueAsString());
} finally {
runner.runDestroy();
}
}
@Test
public void testRenameMapField() throws StageException {
FieldRenamerConfig renameConfig = new FieldRenamerConfig();
// Any field containing a non-word character should be in single quotes
renameConfig.fromFieldExpression = "/first";
renameConfig.toFieldExpression = "/second";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.CONTINUE;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.TO_ERROR;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig), errorHandler);
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.setOnRecordError(OnRecordError.TO_ERROR)
.addOutputLane("a")
.build();
runner.runInit();
Map<String, Field> renameableInnerMap = new HashMap<>();
renameableInnerMap.put("value", Field.create(Field.Type.STRING, "value"));
Map<String, Field> map = new LinkedHashMap<>();
map.put("first", Field.create(renameableInnerMap));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(Field.Type.MAP, map));
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record r = output.getRecords().get("a").get(0);
Assert.assertFalse(r.has("/first"));
Assert.assertFalse(r.has("/first/value"));
Assert.assertTrue(r.has("/second"));
Assert.assertTrue(r.has("/second/value"));
} finally {
runner.runDestroy();
}
}
@Test
public void testNestedKeep2() throws StageException {
/*
* In a deep nested record try to retain arbitrary paths
*/
Record record = createNestedRecord();
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration(
"fields",
ImmutableList.of(
"/USA[0]/SanFrancisco/noe/streets[1][1]/name",
"/USA[1]/SantaMonica/cole/streets[0][1]/name"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0][0]/name"));
Assert.assertEquals(
"d", resultRecord.get("/USA[0]/SanFrancisco/noe/streets[0][0]/name").getValueAsString());
Assert.assertTrue(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][0]/name"));
Assert.assertEquals(
"f", resultRecord.get("/USA[1]/SantaMonica/cole/streets[0][0]/name").getValueAsString());
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[1][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[1][1]/name"));
} finally {
runner.runDestroy();
}
}
@Test
public void testTargetFieldExistsReplace() throws StageException {
// Standard overwrite condition. Source and target fields exist
FieldRenamerConfig renameConfig = new FieldRenamerConfig();
renameConfig.fromFieldExpression = "/existing";
renameConfig.toFieldExpression = "/overwrite";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.CONTINUE;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.REPLACE;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig), errorHandler);
// Test non-existent source with existing target field
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.addOutputLane("a")
.build();
runner.runInit();
try {
Map<String, Field> map = new LinkedHashMap<>();
map.put("existing", Field.create(Field.Type.STRING, "foo"));
map.put("overwrite", Field.create(Field.Type.STRING, "bar"));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Field field = output.getRecords().get("a").get(0).get();
Assert.assertTrue(field.getValue() instanceof Map);
Map<String, Field> result = field.getValueAsMap();
Assert.assertEquals(String.valueOf(result), 1, result.size());
Assert.assertTrue(result.containsKey("overwrite"));
Assert.assertTrue(!result.containsKey("existing"));
Assert.assertEquals("foo", result.get("overwrite").getValue());
} finally {
runner.runDestroy();
}
}
@Test
public void testSourceWithQuotedSubstring() throws StageException {
// source should be processed as quoted string.
FieldRenamerConfig renameConfig1 = new FieldRenamerConfig();
renameConfig1.fromFieldExpression = "/'attr|OrderNum'";
renameConfig1.toFieldExpression = "/theOrderNum";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.CONTINUE;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.REPLACE;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig1), errorHandler);
// Test non-existent source with existing target field
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.addOutputLane("a")
.build();
runner.runInit();
try {
Map<String, Field> map = new LinkedHashMap<>();
map.put("attr|OrderNum", Field.create(Field.Type.STRING, "foo"));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Field field = output.getRecords().get("a").get(0).get();
Assert.assertTrue(field.getValue() instanceof Map);
Map<String, Field> result = field.getValueAsMap();
Assert.assertEquals(String.valueOf(result), 1, result.size());
Assert.assertFalse(result.containsKey("/'attr|OrderNum'"));
Assert.assertFalse(result.containsKey("'attr|OrderNum'"));
Assert.assertFalse(result.containsKey("attr|OrderNum"));
Assert.assertTrue(result.containsKey("theOrderNum"));
} finally {
runner.runDestroy();
}
}
@Test
public void testWildCardKeep2() throws StageException {
/*
* Use wil card in array within array
*/
Record record = createNestedRecord();
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration("fields", ImmutableList.of("/USA[0]/SanFrancisco/noe/streets[0][*]"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0]"));
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[0][0]/name").getValueAsString(), "a");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[0][1]/name").getValueAsString(), "b");
} finally {
runner.runDestroy();
}
/*
* Use wil card in array
*/
record = createNestedRecord();
runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration("fields", ImmutableList.of("/USA[0]/SanFrancisco/noe/streets[*]"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe/streets"));
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[0][0]/name").getValueAsString(), "a");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[0][1]/name").getValueAsString(), "b");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[1][0]/name").getValueAsString(), "c");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[1][1]/name").getValueAsString(), "d");
} finally {
runner.runDestroy();
}
}
@Test
public void testNonExistingSourceAndTargetFields() throws StageException {
// If neither the source or target fields exist, then field renaming is a noop, and should
// succeed
FieldRenamerConfig renameConfig = new FieldRenamerConfig();
renameConfig.fromFieldExpression = "/nonExisting";
renameConfig.toFieldExpression = "/alsoNonExisting";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.CONTINUE;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.TO_ERROR;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig), errorHandler);
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.addOutputLane("a")
.build();
runner.runInit();
try {
Map<String, Field> map = new LinkedHashMap<>();
map.put("name", Field.create(Field.Type.STRING, null));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Field field = output.getRecords().get("a").get(0).get();
Assert.assertTrue(field.getValue() instanceof Map);
Map<String, Field> result = field.getValueAsMap();
Assert.assertEquals(String.valueOf(result), 1, result.size());
Assert.assertTrue(result.containsKey("name"));
Assert.assertEquals(null, result.get("name").getValue());
} finally {
runner.runDestroy();
}
}
@Test
public void testTargetFieldExistsError() throws StageException {
// If overwrite is set to false, overwriting should result in an error
FieldRenamerConfig renameConfig = new FieldRenamerConfig();
renameConfig.fromFieldExpression = "/existing";
renameConfig.toFieldExpression = "/overwrite";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.CONTINUE;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.TO_ERROR;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig), errorHandler);
// Test non-existent source with existing target field
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.setOnRecordError(OnRecordError.TO_ERROR)
.addOutputLane("a")
.build();
runner.runInit();
try {
Map<String, Field> map = new LinkedHashMap<>();
map.put("existing", Field.create(Field.Type.STRING, "foo"));
map.put("overwrite", Field.create(Field.Type.STRING, "bar"));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(0, output.getRecords().get("a").size());
Assert.assertEquals(1, runner.getErrorRecords().size());
} finally {
runner.runDestroy();
}
}
@Test
public void testSingleRecordMap() throws Exception {
List<JdbcFieldColumnMapping> columnMappings =
ImmutableList.of(new JdbcFieldColumnMapping("P_ID", "/p_id"));
JdbcLookupDProcessor processor = new JdbcLookupDProcessor();
processor.hikariConfigBean = createConfigBean(h2ConnectionString, username, password);
ProcessorRunner processorRunner =
new ProcessorRunner.Builder(JdbcLookupDProcessor.class, processor)
.addConfiguration("query", mapQuery)
.addConfiguration("columnMappings", columnMappings)
.addConfiguration("maxClobSize", 1000)
.addConfiguration("maxBlobSize", 1000)
.addOutputLane("lane")
.build();
Record record = RecordCreator.create();
LinkedHashMap<String, Field> fields = new LinkedHashMap<>();
fields.put("first_name", Field.create("Adam"));
fields.put("last_name", Field.create("Kunicki"));
record.set(Field.create(fields));
List<Record> singleRecord = ImmutableList.of(record);
processorRunner.runInit();
try {
StageRunner.Output output = processorRunner.runProcess(singleRecord);
Assert.assertEquals(1, output.getRecords().get("lane").size());
record = output.getRecords().get("lane").get(0);
Assert.assertNotEquals(null, record.get("/p_id"));
Assert.assertEquals(1, record.get("/p_id").getValueAsInteger());
} finally {
processorRunner.runDestroy();
}
}
@Test
public void testUnreachableFields() throws Exception {
FieldRenamerConfig renameConfig = new FieldRenamerConfig();
renameConfig.fromFieldExpression = "/a";
renameConfig.toFieldExpression = "/b/c/d";
FieldRenamerProcessorErrorHandler errorHandler = new FieldRenamerProcessorErrorHandler();
errorHandler.nonExistingFromFieldHandling = OnStagePreConditionFailure.TO_ERROR;
errorHandler.multipleFromFieldsMatching = OnStagePreConditionFailure.TO_ERROR;
errorHandler.existingToFieldHandling = ExistingToFieldHandling.REPLACE;
FieldRenamerProcessor processor =
new FieldRenamerProcessor(ImmutableList.of(renameConfig), errorHandler);
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldRenamerDProcessor.class, processor)
.addOutputLane("a")
.setOnRecordError(OnRecordError.TO_ERROR)
.build();
runner.runInit();
try {
Map<String, Field> map = new LinkedHashMap<>();
map.put("a", Field.create(123));
Record record = RecordCreator.create("s", "s:1");
record.set(Field.create(map));
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(0, output.getRecords().get("a").size());
Assert.assertEquals(1, runner.getErrorRecords().size());
Record errorRecord = runner.getErrorRecords().get(0);
Assert.assertEquals(Errors.FIELD_RENAMER_04.name(), errorRecord.getHeader().getErrorCode());
} finally {
runner.runDestroy();
}
}
@Test
public void testNestedKeep4() throws StageException {
Record record = createNestedRecord();
/*
* Keep non existing nested path "/USA[0]/SanFrancisco/cole".
* Only objects upto /USA[0]/SanFrancisco should exist
*/
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration(
"fields",
ImmutableList.of(
"/USA[0]/SanFrancisco/cole/streets[0][0]/name",
"/USA[0]/SanFrancisco/cole/streets[1][0]/name"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco"));
Assert.assertEquals(0, resultRecord.get("/USA[0]/SanFrancisco").getValueAsMap().size());
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0][1]/name"));
} finally {
runner.runDestroy();
}
}
@Test
public void testNestedKeep1() throws StageException {
/*
* In a deep nested field path try to retain the second elements of an array within an array
*/
Record record = createNestedRecord();
// Keep only second elements of array within array "/USA[0]/SanFrancisco/noe/streets"
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration(
"fields",
ImmutableList.of(
"/USA[0]/SanFrancisco/noe/streets[0][1]/name",
"/USA[0]/SanFrancisco/noe/streets[1][1]/name"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
// Note that since the first element was removed, the second element is the new first element
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0][0]/name"));
Assert.assertEquals(
"b", resultRecord.get("/USA[0]/SanFrancisco/noe/streets[0][0]/name").getValueAsString());
// Note that since the first element was removed, the second element is the new first element
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[1][0]/name"));
Assert.assertEquals(
"d", resultRecord.get("/USA[0]/SanFrancisco/noe/streets[1][0]/name").getValueAsString());
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[1][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0][1]/name"));
} finally {
runner.runDestroy();
}
/*
* In a deep nested field path try to retain the first elements of an array within an array
*/
record = createNestedRecord();
// Keep only first elements of array within array "/USA[0]/SanFrancisco/noe/streets"
runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration(
"fields",
ImmutableList.of(
"/USA[0]/SanFrancisco/noe/streets[0][0]/name",
"/USA[0]/SanFrancisco/noe/streets[1][0]/name"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0][0]/name"));
Assert.assertEquals(
"a", resultRecord.get("/USA[0]/SanFrancisco/noe/streets[0][0]/name").getValueAsString());
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[1][0]/name"));
Assert.assertEquals(
"c", resultRecord.get("/USA[0]/SanFrancisco/noe/streets[1][0]/name").getValueAsString());
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[1][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0][1]/name"));
} finally {
runner.runDestroy();
}
}
@Test
public void testWildCardKeep3() throws StageException {
/*
* Use wil card in map and array
*/
Record record = createNestedRecord();
ProcessorRunner runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration(
"fields", ImmutableList.of("/USA[0]/SanFrancisco/*/streets[*][1]/name"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0][0]/name"));
Assert.assertEquals(
"b", resultRecord.get("/USA[0]/SanFrancisco/noe/streets[0][0]/name").getValueAsString());
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[0][1]/name"));
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[1][0]/name"));
Assert.assertEquals(
"d", resultRecord.get("/USA[0]/SanFrancisco/noe/streets[1][0]/name").getValueAsString());
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/noe/streets[1][1]/name"));
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[0]/SanFrancisco/folsom/streets[0][1]/name"));
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][1]/name"));
} finally {
runner.runDestroy();
}
/*
* Use wil card in map. Make sure the entire tree of elements is preserved
*/
record = createNestedRecord();
runner =
new ProcessorRunner.Builder(FieldFilterDProcessor.class)
.addConfiguration("fields", ImmutableList.of("/USA[0]/SanFrancisco/*"))
.addConfiguration("filterOperation", FilterOperation.KEEP)
.addOutputLane("a")
.build();
runner.runInit();
try {
StageRunner.Output output = runner.runProcess(ImmutableList.of(record));
Assert.assertEquals(1, output.getRecords().get("a").size());
Record resultRecord = output.getRecords().get("a").get(0);
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/noe"));
Assert.assertTrue(resultRecord.has("/USA[0]/SanFrancisco/folsom"));
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[0][0]/name").getValueAsString(), "a");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[0][1]/name").getValueAsString(), "b");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[1][0]/name").getValueAsString(), "c");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/noe/streets[1][1]/name").getValueAsString(), "d");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/folsom/streets[0][0]/name").getValueAsString(), "g");
Assert.assertEquals(
record.get("/USA[0]/SanFrancisco/folsom/streets[0][1]/name").getValueAsString(), "h");
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][0]/name"));
Assert.assertFalse(resultRecord.has("/USA[1]/SantaMonica/cole/streets[0][1]/name"));
} finally {
runner.runDestroy();
}
}
@Test(timeout = 30000)
public void testProduceAvroData() throws Exception {
SourceRunner sourceRunner =
new SourceRunner.Builder(ClusterHdfsDSource.class)
.addOutputLane("lane")
.setExecutionMode(ExecutionMode.CLUSTER_BATCH)
.addConfiguration("hdfsUri", miniDFS.getURI().toString())
.addConfiguration("hdfsDirLocations", Arrays.asList(dir.toUri().getPath()))
.addConfiguration("recursive", false)
.addConfiguration("hdfsConfigs", new HashMap<String, String>())
.addConfiguration("dataFormat", DataFormat.AVRO)
.addConfiguration("csvFileFormat", CsvMode.CSV)
.addConfiguration("csvHeader", CsvHeader.WITH_HEADER)
.addConfiguration("csvMaxObjectLen", 4096)
.addConfiguration("textMaxLineLen", 1024)
.addConfiguration("produceSingleRecordPerMessage", false)
.addConfiguration("regex", null)
.addConfiguration("grokPatternDefinition", null)
.addConfiguration("enableLog4jCustomLogFormat", false)
.addConfiguration("customLogFormat", null)
.addConfiguration("fieldPathsToGroupName", null)
.addConfiguration("log4jCustomLogFormat", null)
.addConfiguration("grokPattern", null)
.addConfiguration("hdfsKerberos", false)
.addConfiguration("hdfsConfDir", hadoopConfDir)
.setResourcesDir(resourcesDir)
.build();
sourceRunner.runInit();
List<Map.Entry> list = new ArrayList<>();
list.add(
new Pair(
"path::" + "1" + "::1",
createAvroData("a", 30, ImmutableList.of("*****@*****.**", "*****@*****.**"))));
list.add(
new Pair(
"path::" + "1" + "::2",
createAvroData("b", 40, ImmutableList.of("*****@*****.**", "*****@*****.**"))));
Thread th = createThreadForAddingBatch(sourceRunner, list);
try {
StageRunner.Output output = sourceRunner.runProduce(null, 5);
String newOffset = output.getNewOffset();
Assert.assertEquals("path::" + "1::2", newOffset);
List<Record> records = output.getRecords().get("lane");
Assert.assertEquals(2, records.size());
Record record = records.get(0);
Assert.assertTrue(record.has("/name"));
Assert.assertEquals("a", record.get("/name").getValueAsString());
Assert.assertTrue(record.has("/age"));
Assert.assertEquals(30, record.get("/age").getValueAsInteger());
Assert.assertTrue(record.has("/emails"));
Assert.assertTrue(record.get("/emails").getValueAsList() instanceof List);
List<Field> emails = record.get("/emails").getValueAsList();
Assert.assertEquals(2, emails.size());
Assert.assertEquals("*****@*****.**", emails.get(0).getValueAsString());
Assert.assertEquals("*****@*****.**", emails.get(1).getValueAsString());
record = records.get(1);
Assert.assertTrue(record.has("/name"));
Assert.assertEquals("b", record.get("/name").getValueAsString());
Assert.assertTrue(record.has("/age"));
Assert.assertEquals(40, record.get("/age").getValueAsInteger());
Assert.assertTrue(record.has("/emails"));
Assert.assertTrue(record.get("/emails").getValueAsList() instanceof List);
emails = record.get("/emails").getValueAsList();
Assert.assertEquals(2, emails.size());
Assert.assertEquals("*****@*****.**", emails.get(0).getValueAsString());
Assert.assertEquals("*****@*****.**", emails.get(1).getValueAsString());
} finally {
th.interrupt();
}
}