@BeforeClass
public static void before() throws Exception {
final String filePath =
TestUtils.getFileFromResourceUrl(
DictionariesTest.class.getClassLoader().getResource(AVRO_DATA));
if (INDEX_DIR.exists()) {
FileUtils.deleteQuietly(INDEX_DIR);
}
final SegmentGeneratorConfig config =
SegmentTestUtils.getSegmentGenSpecWithSchemAndProjectedColumns(
new File(filePath), INDEX_DIR, "time_day", TimeUnit.DAYS, "test");
final SegmentIndexCreationDriver driver = SegmentCreationDriverFactory.get(null);
driver.init(config);
driver.build();
final Schema schema = AvroUtils.extractSchemaFromAvro(new File(filePath));
final DataFileStream<GenericRecord> avroReader = AvroUtils.getAvroReader(new File(filePath));
final org.apache.avro.Schema avroSchema = avroReader.getSchema();
final String[] columns = new String[avroSchema.getFields().size()];
int i = 0;
for (final Field f : avroSchema.getFields()) {
columns[i] = f.name();
i++;
}
uniqueEntries = new HashMap<String, Set<Object>>();
for (final String column : columns) {
uniqueEntries.put(column, new HashSet<Object>());
}
while (avroReader.hasNext()) {
final GenericRecord rec = avroReader.next();
for (final String column : columns) {
Object val = rec.get(column);
if (val instanceof Utf8) {
val = ((Utf8) val).toString();
}
uniqueEntries
.get(column)
.add(getAppropriateType(schema.getFieldSpecFor(column).getDataType(), val));
}
}
}
@Test
public void testAvroNativeJson() throws IOException {
AvroNativeFileOutputFormat format = new AvroNativeFileOutputFormat();
ByteArrayOutputStream sos = new ByteArrayOutputStream();
format.format(sos, e);
format.close();
byte[] bytes = sos.toByteArray();
ReflectData reflectData = ReflectData.get();
Schema schema = reflectData.getSchema(EventImpl.class);
ReflectDatumReader<EventImpl> dr = new ReflectDatumReader<EventImpl>(schema);
ByteArrayInputStream bais = new ByteArrayInputStream(bytes);
DataFileStream<EventImpl> dec = new DataFileStream<EventImpl>(bais, dr);
Event er = dec.next();
assertEquals(e.getHost(), er.getHost());
assertEquals(e.getNanos(), er.getNanos());
assertEquals(e.getPriority(), er.getPriority());
assertTrue(Arrays.equals(e.getBody(), er.getBody()));
}
public void generateSimpleAggregationOnSingleColumnFilters() throws IOException {
final Map<String, Map<Object, Integer>> cardinalityCountsMap =
new HashMap<String, Map<Object, Integer>>();
final Map<String, Map<Object, Map<String, Double>>> sumMap =
new HashMap<String, Map<Object, Map<String, Double>>>();
// here string key is columnName:columnValue:MetricName:GroupColumnName:groupKey:metricValue
final Map<String, Map<Object, Double>> sumGroupBy = new HashMap<String, Map<Object, Double>>();
aggregationQueries = new ArrayList<AvroQueryGenerator.TestSimpleAggreationQuery>();
groupByQueries = new ArrayList<AvroQueryGenerator.TestGroupByAggreationQuery>();
for (final Field f : schema.getFields()) {
final String fieldName = f.name();
if (dimensions.contains(fieldName) || metrics.contains(fieldName) || time.equals(fieldName)) {
isSingleValueMap.put(fieldName, isSingleValueField(f));
dataTypeMap.put(fieldName, getColumnType(f));
if (!metrics.contains(fieldName)) {
cardinalityCountsMap.put(fieldName, new HashMap<Object, Integer>());
}
}
}
for (final String column : cardinalityCountsMap.keySet()) {
sumMap.put(column, new HashMap<Object, Map<String, Double>>());
}
// here string key is columnName:columnValue:MetricName:GroupColumnName:groupKey:metricValue
while (dataStream.hasNext()) {
final GenericRecord record = dataStream.next();
for (final String column : cardinalityCountsMap.keySet()) {
Object value = record.get(column);
if (value == null) {
switch (schema.getField(column).schema().getType()) {
case INT:
value = 0;
break;
case FLOAT:
value = 0F;
break;
case LONG:
value = 0L;
break;
case DOUBLE:
value = 0D;
break;
case STRING:
case BOOLEAN:
value = "null";
break;
}
}
if (value instanceof Utf8) {
value = ((Utf8) value).toString();
}
if (value instanceof Array) {
continue;
}
// here string key is columnName:columnValue:MetricName:GroupColumnName:groupKey:metricValue
for (final String metricName : metrics) {
final String groupbyKeyBase = column + ":" + record.get(column) + ":" + metricName;
int dimCounter = 1;
for (final String dim : cardinalityCountsMap.keySet()) {
if (!dim.equals(column)) {
dimCounter++;
final String groupbyKey = groupbyKeyBase + ":" + dim;
if (sumGroupBy.containsKey(groupbyKey)) {
if (sumGroupBy.get(groupbyKey).containsKey(record.get(dim))) {
sumGroupBy
.get(groupbyKey)
.put(
record.get(dim),
getAppropriateNumberType(
metricName,
record.get(metricName),
sumGroupBy.get(groupbyKey).get(record.get(dim))));
} else {
sumGroupBy
.get(groupbyKey)
.put(record.get(dim), Double.parseDouble(record.get(metricName).toString()));
}
} else {
sumGroupBy.put(groupbyKey, new HashMap<Object, Double>());
sumGroupBy
.get(groupbyKey)
.put(record.get(dim), Double.parseDouble(record.get(metricName).toString()));
}
}
if (dimCounter == 4) {
break;
}
}
}
if (cardinalityCountsMap.get(column).containsKey(value)) {
cardinalityCountsMap
.get(column)
.put(value, cardinalityCountsMap.get(column).get(value) + 1);
} else {
cardinalityCountsMap.get(column).put(value, 1);
}
if (!sumMap.get(column).containsKey(value)) {
sumMap.get(column).put(value, new HashMap<String, Double>());
}
for (final String metric : metrics) {
if (!sumMap.get(column).get(value).containsKey(metric)) {
sumMap
.get(column)
.get(value)
.put(metric, getAppropriateNumberType(metric, record.get(metric), 0D));
} else {
sumMap
.get(column)
.get(value)
.put(
metric,
getAppropriateNumberType(
metric, record.get(metric), sumMap.get(column).get(value).get(metric)));
}
}
// here string key is columnName:columnValue:MetricName:GroupColumnName:groupKey:metricValue
}
}
dataStream.close();
if (!isRealtimeSegment) {
for (final String column : cardinalityCountsMap.keySet()) {
for (final Object entry : cardinalityCountsMap.get(column).keySet()) {
final StringBuilder bld = new StringBuilder();
bld.append("select count(*) from ");
bld.append(resourceName);
bld.append(" where ");
bld.append(column);
bld.append("=");
bld.append("'");
bld.append(entry);
bld.append("'");
bld.append(" ");
bld.append("limit 0");
String queryString = bld.toString();
if (!queryString.contains("null")) {
aggregationQueries.add(
new TestSimpleAggreationQuery(
queryString, new Double(cardinalityCountsMap.get(column).get(entry))));
}
}
}
}
for (final String column : sumMap.keySet()) {
for (final Object value : sumMap.get(column).keySet()) {
for (final String metric : sumMap.get(column).get(value).keySet()) {
final StringBuilder bld = new StringBuilder();
bld.append("select sum('" + metric + "') from ");
bld.append(resourceName);
bld.append(" where ");
bld.append(column);
bld.append("=");
bld.append("'");
bld.append(value);
bld.append("'");
bld.append(" ");
bld.append("limit 0");
String queryString = bld.toString();
if (!queryString.contains("null")) {
aggregationQueries.add(
new TestSimpleAggreationQuery(
bld.toString(), sumMap.get(column).get(value).get(metric)));
}
}
}
}
for (final String groupKey : sumGroupBy.keySet()) {
final String columnName = groupKey.split(":")[0];
final String columnValue = groupKey.split(":")[1];
final String metricColumn = groupKey.split(":")[2];
final String groupByColumnName = groupKey.split(":")[3];
final StringBuilder bld = new StringBuilder();
bld.append("select sum('" + metricColumn + "') from ");
bld.append(resourceName);
bld.append(" where ");
bld.append(columnName);
bld.append("=");
bld.append("'");
bld.append(columnValue);
bld.append("'");
bld.append(" ");
bld.append(" group by ");
bld.append(groupByColumnName);
bld.append(" top 10 ");
bld.append("limit 0");
String queryString = bld.toString();
if (!queryString.contains("null")) {
groupByQueries.add(
new TestGroupByAggreationQuery(bld.toString(), sumGroupBy.get(groupKey)));
}
}
}
public void init() throws FileNotFoundException, IOException {
dataStream =
new DataFileStream<GenericRecord>(
new FileInputStream(avroFile), new GenericDatumReader<GenericRecord>());
schema = dataStream.getSchema();
}