@Test
public void testNestedTypes() {
MessageType schema = MessageTypeParser.parseMessageType(Paper.schema.toString());
Type type = schema.getType("Links", "Backward");
assertEquals(PrimitiveTypeName.INT64, type.asPrimitiveType().getPrimitiveTypeName());
assertEquals(0, schema.getMaxRepetitionLevel("DocId"));
assertEquals(1, schema.getMaxRepetitionLevel("Name"));
assertEquals(2, schema.getMaxRepetitionLevel("Name", "Language"));
assertEquals(0, schema.getMaxDefinitionLevel("DocId"));
assertEquals(1, schema.getMaxDefinitionLevel("Links"));
assertEquals(2, schema.getMaxDefinitionLevel("Links", "Backward"));
}
@Test
public void testIDs() throws Exception {
MessageType schema =
new MessageType(
"test",
new PrimitiveType(REQUIRED, BINARY, "foo").withId(4),
new GroupType(REQUIRED, "bar", new PrimitiveType(REQUIRED, BINARY, "baz").withId(3))
.withId(8));
MessageType schema2 = MessageTypeParser.parseMessageType(schema.toString());
assertEquals(schema, schema2);
assertEquals(schema.toString(), schema2.toString());
}
/**
* read GPDBWritable from gpdb and then write it to hdfs
*
* @throws Exception when something goes wrong
*/
public void doWrite() throws IOException {
// if there is no schema provided by user, we will read schema later
ParquetWriter<Group> dataFileWriter = null;
DataInputStream dis = new DataInputStream(System.in);
try {
MessageType schema = null;
SimpleGroupFactory groupFactory = null;
// read table structure info and auto-gen avro schema
schema = autoGenSchema(dis);
// int total = dis.readInt();//skip the original 4 byte VARHDSZ
if (parquetSchemaFile != null) {
// if user give us a schema file, read schema from it
String schemaString = readSchemaFile(parquetSchemaFile);
schema = MessageTypeParser.parseMessageType(schemaString);
}
GroupWriteSupport.setSchema(schema, conf);
CompressionCodecName codecName = CompressionCodecName.UNCOMPRESSED;
if (isCompressed) {
if (compressCodec.equals(LZO_COMPRESS)) {
codecName = CompressionCodecName.LZO;
} else if (compressCodec.equals(SNAPPY_COMPRESS)) {
codecName = CompressionCodecName.SNAPPY;
} else if (compressCodec.equals(GZIP_COMPRESS)) {
codecName = CompressionCodecName.GZIP;
} else {
throw new IOException("compression method not support, codec:" + compressCodec);
}
}
dataFileWriter =
new ParquetWriter<Group>(
new Path(outputPath),
new GroupWriteSupport(),
codecName,
rowGroupSize,
pageSize,
dicPageSize,
dicEnable,
false,
parquetVersion,
conf);
groupFactory = new SimpleGroupFactory(schema);
while (true) {
GPDBWritable gw = new GPDBWritable();
gw.readFields(dis);
Group pqGroup = groupFactory.newGroup();
fillRecord(pqGroup, gw, schema);
dataFileWriter.write(pqGroup);
}
} catch (EOFException e) {
// this is ok, read the end of input stream, keep error msg for testing
// e.printStackTrace();
} finally {
if (dataFileWriter != null) {
dataFileWriter.close();
}
dis.close();
}
}
@Test
public void test() throws Exception {
MessageType schema = MessageTypeParser.parseMessageType(Paper.schema.toString());
assertEquals(Paper.schema, schema);
assertEquals(schema.toString(), Paper.schema.toString());
}
