@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));
}
}
}
public void init() throws FileNotFoundException, IOException {
dataStream =
new DataFileStream<GenericRecord>(
new FileInputStream(avroFile), new GenericDatumReader<GenericRecord>());
schema = dataStream.getSchema();
}
