@Test
public void testCollections() throws Exception {
Collection<String> j = Lists.newArrayList();
j.add("a");
j.add("b");
Schema collectionSchema = Schema.createArray(Avros.strings().getSchema());
GenericData.Array<Utf8> w = new GenericData.Array<Utf8>(2, collectionSchema);
w.add(new Utf8("a"));
w.add(new Utf8("b"));
testInputOutputFn(Avros.collections(Avros.strings()), j, w);
}
/** {@inheritDoc} */
@Override
public Array<CharSequence> convert(List<String> recommendationList) {
List<CharSequence> recommendationArray = new ArrayList<CharSequence>();
;
for (String s : recommendationList) {
recommendationArray.add(s);
}
Array<CharSequence> recomendationArray =
new Array<CharSequence>(
Schema.createArray(Schema.create(Schema.Type.STRING)), recommendationArray);
return recomendationArray;
}
public static Schema getAvroSchema(FieldType type) {
switch (type) {
case STRING:
return Schema.create(Schema.Type.STRING);
case BINARY:
return Schema.create(Schema.Type.BYTES);
case DOUBLE:
case DECIMAL:
return Schema.create(Schema.Type.DOUBLE);
case BOOLEAN:
return Schema.create(Schema.Type.BOOLEAN);
case DATE:
case TIME:
case INTEGER:
return Schema.create(Schema.Type.INT);
case LONG:
case TIMESTAMP:
return Schema.create(Schema.Type.LONG);
default:
if (type.isMap()) {
Schema union =
Schema.createUnion(
ImmutableList.of(
Schema.create(Schema.Type.NULL), getAvroSchema(type.getMapValueType())));
return Schema.createMap(union);
}
if (type.isArray()) {
Schema union =
Schema.createUnion(
ImmutableList.of(
Schema.create(Schema.Type.NULL), getAvroSchema(type.getArrayElementType())));
return Schema.createArray(union);
}
throw new IllegalStateException();
}
}
static {
Schemas.STRING_LANG.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("label", Schemas.STRING, null, null),
new Schema.Field("language", Schemas.STRING, null, null)));
Schemas.SHORT.setFields(
ImmutableList.<Schema.Field>of(new Schema.Field("short", Schemas.INT, null, null)));
Schemas.BYTE.setFields(
ImmutableList.<Schema.Field>of(new Schema.Field("byte", Schemas.INT, null, null)));
Schemas.BIGINTEGER.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("biginteger", Schemas.STRING, null, null)));
Schemas.BIGDECIMAL.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("bigdecimal", Schemas.STRING, null, null)));
Schemas.PLAIN_IDENTIFIER.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("identifier", Schemas.STRING, null, null)));
Schemas.COMPRESSED_IDENTIFIER.setFields(
ImmutableList.<Schema.Field>of(new Schema.Field("identifier", Schemas.INT, null, null)));
Schemas.CALENDAR.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("timezone", Schemas.INT, null, null),
new Schema.Field("timestamp", Schemas.LONG, null, null)));
Schemas.STATEMENT.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("subject", Schemas.IDENTIFIER, null, null),
new Schema.Field("predicate", Schemas.IDENTIFIER, null, null),
new Schema.Field(
"object",
Schema.createUnion(
ImmutableList.<Schema>of(
Schemas.BOOLEAN,
Schemas.STRING,
Schemas.STRING_LANG,
Schemas.LONG,
Schemas.INT,
Schemas.SHORT,
Schemas.BYTE,
Schemas.DOUBLE,
Schemas.FLOAT,
Schemas.BIGINTEGER,
Schemas.BIGDECIMAL,
Schemas.CALENDAR,
Schemas.PLAIN_IDENTIFIER,
Schemas.COMPRESSED_IDENTIFIER)),
null,
null), //
new Schema.Field("context", Schemas.IDENTIFIER, null, null)));
Schemas.PROPERTY.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("propertyURI", Schemas.COMPRESSED_IDENTIFIER, null, null),
new Schema.Field(
"propertyValue",
Schema.createUnion(
ImmutableList.<Schema>of(
Schemas.BOOLEAN,
Schemas.STRING,
Schemas.STRING_LANG,
Schemas.LONG,
Schemas.INT,
Schemas.SHORT,
Schemas.BYTE,
Schemas.DOUBLE,
Schemas.FLOAT,
Schemas.BIGINTEGER,
Schemas.BIGDECIMAL,
Schemas.CALENDAR,
Schemas.PLAIN_IDENTIFIER,
Schemas.COMPRESSED_IDENTIFIER,
Schemas.STATEMENT,
Schemas.RECORD,
Schemas.LIST)),
null,
null)));
Schemas.RECORD.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field(
"id",
Schema.createUnion(
ImmutableList.<Schema>of(
Schemas.NULL, Schemas.PLAIN_IDENTIFIER, Schemas.COMPRESSED_IDENTIFIER)),
null,
null), //
new Schema.Field("properties", Schema.createArray(Schemas.PROPERTY), null, null)));
}
private static final class Schemas {
/** The namespace for KS-specific AVRO schemas. */
public static final String NAMESPACE = "eu.fbk.knowledgestore";
/** AVRO schema for NULL. */
public static final Schema NULL = Schema.create(Schema.Type.NULL);
/** AVRO schema for boolean literals. */
public static final Schema BOOLEAN = Schema.create(Schema.Type.BOOLEAN);
/** AVRO schema for string literals. */
public static final Schema STRING = Schema.create(Schema.Type.STRING);
/** AVRO schema for string literals with a language. */
public static final Schema STRING_LANG =
Schema.createRecord("stringlang", null, Schemas.NAMESPACE, false);
/** AVRO schema for long literals. */
public static final Schema LONG = Schema.create(Schema.Type.LONG);
/** AVRO schema for int literals. */
public static final Schema INT = Schema.create(Schema.Type.INT);
/** AVRO schema for short literals. */
public static final Schema SHORT = Schema.createRecord("short", null, Schemas.NAMESPACE, false);
/** AVRO schema for byte literals. */
public static final Schema BYTE = Schema.createRecord("byte", null, Schemas.NAMESPACE, false);
/** AVRO schema for double literals. */
public static final Schema DOUBLE = Schema.create(Schema.Type.DOUBLE);
/** AVRO schema for float literals. */
public static final Schema FLOAT = Schema.create(Schema.Type.FLOAT);
/** AVRO schema for big integer literals. */
public static final Schema BIGINTEGER =
Schema.createRecord("biginteger", null, Schemas.NAMESPACE, false);
/** AVRO schema for big decimal literals. */
public static final Schema BIGDECIMAL =
Schema.createRecord("bigdecimal", null, Schemas.NAMESPACE, false);
/** AVRO schema for non-compressed IDs (URIs, BNodes). */
public static final Schema PLAIN_IDENTIFIER =
Schema //
.createRecord("plainidentifier", null, Schemas.NAMESPACE, false);
/** AVRO schema for compressed ID (URIs, BNodes). */
public static final Schema COMPRESSED_IDENTIFIER =
Schema //
.createRecord("compressedidentifier", null, Schemas.NAMESPACE, false);
/** AVRO schema for any ID (URIs, BNodes). */
public static final Schema IDENTIFIER =
Schema.createUnion(ImmutableList.<Schema>of(PLAIN_IDENTIFIER, COMPRESSED_IDENTIFIER));
/** AVRO schema for calendar literals. */
public static final Schema CALENDAR =
Schema.createRecord("calendar", null, Schemas.NAMESPACE, false);
/** AVRO schema for RDF statements. */
public static final Schema STATEMENT =
Schema.createRecord("statement", null, Schemas.NAMESPACE, false);
/** AVRO schema for record nodes ({@code Record}). */
public static final Schema RECORD =
Schema.createRecord("struct", null, Schemas.NAMESPACE, false);
/** AVRO schema for generic data model nodes. */
public static final Schema NODE =
Schema.createUnion(
ImmutableList.<Schema>of(
Schemas.BOOLEAN,
Schemas.STRING,
Schemas.STRING_LANG,
Schemas.LONG,
Schemas.INT,
Schemas.SHORT,
Schemas.BYTE,
Schemas.DOUBLE,
Schemas.FLOAT,
Schemas.BIGINTEGER,
Schemas.BIGDECIMAL,
Schemas.PLAIN_IDENTIFIER,
Schemas.COMPRESSED_IDENTIFIER,
Schemas.CALENDAR,
Schemas.STATEMENT,
Schemas.RECORD));
/** AVRO schema for lists of nodes. */
public static final Schema LIST = Schema.createArray(Schemas.NODE);
/** AVRO schema for properties of a record node. */
public static final Schema PROPERTY =
Schema.createRecord("property", null, Schemas.NAMESPACE, false);
private Schemas() {}
static {
Schemas.STRING_LANG.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("label", Schemas.STRING, null, null),
new Schema.Field("language", Schemas.STRING, null, null)));
Schemas.SHORT.setFields(
ImmutableList.<Schema.Field>of(new Schema.Field("short", Schemas.INT, null, null)));
Schemas.BYTE.setFields(
ImmutableList.<Schema.Field>of(new Schema.Field("byte", Schemas.INT, null, null)));
Schemas.BIGINTEGER.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("biginteger", Schemas.STRING, null, null)));
Schemas.BIGDECIMAL.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("bigdecimal", Schemas.STRING, null, null)));
Schemas.PLAIN_IDENTIFIER.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("identifier", Schemas.STRING, null, null)));
Schemas.COMPRESSED_IDENTIFIER.setFields(
ImmutableList.<Schema.Field>of(new Schema.Field("identifier", Schemas.INT, null, null)));
Schemas.CALENDAR.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("timezone", Schemas.INT, null, null),
new Schema.Field("timestamp", Schemas.LONG, null, null)));
Schemas.STATEMENT.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("subject", Schemas.IDENTIFIER, null, null),
new Schema.Field("predicate", Schemas.IDENTIFIER, null, null),
new Schema.Field(
"object",
Schema.createUnion(
ImmutableList.<Schema>of(
Schemas.BOOLEAN,
Schemas.STRING,
Schemas.STRING_LANG,
Schemas.LONG,
Schemas.INT,
Schemas.SHORT,
Schemas.BYTE,
Schemas.DOUBLE,
Schemas.FLOAT,
Schemas.BIGINTEGER,
Schemas.BIGDECIMAL,
Schemas.CALENDAR,
Schemas.PLAIN_IDENTIFIER,
Schemas.COMPRESSED_IDENTIFIER)),
null,
null), //
new Schema.Field("context", Schemas.IDENTIFIER, null, null)));
Schemas.PROPERTY.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field("propertyURI", Schemas.COMPRESSED_IDENTIFIER, null, null),
new Schema.Field(
"propertyValue",
Schema.createUnion(
ImmutableList.<Schema>of(
Schemas.BOOLEAN,
Schemas.STRING,
Schemas.STRING_LANG,
Schemas.LONG,
Schemas.INT,
Schemas.SHORT,
Schemas.BYTE,
Schemas.DOUBLE,
Schemas.FLOAT,
Schemas.BIGINTEGER,
Schemas.BIGDECIMAL,
Schemas.CALENDAR,
Schemas.PLAIN_IDENTIFIER,
Schemas.COMPRESSED_IDENTIFIER,
Schemas.STATEMENT,
Schemas.RECORD,
Schemas.LIST)),
null,
null)));
Schemas.RECORD.setFields(
ImmutableList.<Schema.Field>of(
new Schema.Field(
"id",
Schema.createUnion(
ImmutableList.<Schema>of(
Schemas.NULL, Schemas.PLAIN_IDENTIFIER, Schemas.COMPRESSED_IDENTIFIER)),
null,
null), //
new Schema.Field("properties", Schema.createArray(Schemas.PROPERTY), null, null)));
}
}
@Test
public void testAllUsingDefaultAvroSchema() throws Exception {
File tmp = File.createTempFile(getClass().getSimpleName(), ".tmp");
tmp.deleteOnExit();
tmp.delete();
Path file = new Path(tmp.getPath());
// write file using Parquet APIs
ParquetWriter<Map<String, Object>> parquetWriter =
new ParquetWriter<Map<String, Object>>(
file,
new WriteSupport<Map<String, Object>>() {
private RecordConsumer recordConsumer;
@Override
public WriteContext init(Configuration configuration) {
return new WriteContext(
MessageTypeParser.parseMessageType(TestAvroSchemaConverter.ALL_PARQUET_SCHEMA),
new HashMap<String, String>());
}
@Override
public void prepareForWrite(RecordConsumer recordConsumer) {
this.recordConsumer = recordConsumer;
}
@Override
public void write(Map<String, Object> record) {
recordConsumer.startMessage();
int index = 0;
recordConsumer.startField("myboolean", index);
recordConsumer.addBoolean((Boolean) record.get("myboolean"));
recordConsumer.endField("myboolean", index++);
recordConsumer.startField("myint", index);
recordConsumer.addInteger((Integer) record.get("myint"));
recordConsumer.endField("myint", index++);
recordConsumer.startField("mylong", index);
recordConsumer.addLong((Long) record.get("mylong"));
recordConsumer.endField("mylong", index++);
recordConsumer.startField("myfloat", index);
recordConsumer.addFloat((Float) record.get("myfloat"));
recordConsumer.endField("myfloat", index++);
recordConsumer.startField("mydouble", index);
recordConsumer.addDouble((Double) record.get("mydouble"));
recordConsumer.endField("mydouble", index++);
recordConsumer.startField("mybytes", index);
recordConsumer.addBinary(
Binary.fromReusedByteBuffer((ByteBuffer) record.get("mybytes")));
recordConsumer.endField("mybytes", index++);
recordConsumer.startField("mystring", index);
recordConsumer.addBinary(Binary.fromString((String) record.get("mystring")));
recordConsumer.endField("mystring", index++);
recordConsumer.startField("mynestedrecord", index);
recordConsumer.startGroup();
recordConsumer.startField("mynestedint", 0);
recordConsumer.addInteger((Integer) record.get("mynestedint"));
recordConsumer.endField("mynestedint", 0);
recordConsumer.endGroup();
recordConsumer.endField("mynestedrecord", index++);
recordConsumer.startField("myenum", index);
recordConsumer.addBinary(Binary.fromString((String) record.get("myenum")));
recordConsumer.endField("myenum", index++);
recordConsumer.startField("myarray", index);
recordConsumer.startGroup();
recordConsumer.startField("array", 0);
for (int val : (int[]) record.get("myarray")) {
recordConsumer.addInteger(val);
}
recordConsumer.endField("array", 0);
recordConsumer.endGroup();
recordConsumer.endField("myarray", index++);
recordConsumer.startField("myoptionalarray", index);
recordConsumer.startGroup();
recordConsumer.startField("array", 0);
for (int val : (int[]) record.get("myoptionalarray")) {
recordConsumer.addInteger(val);
}
recordConsumer.endField("array", 0);
recordConsumer.endGroup();
recordConsumer.endField("myoptionalarray", index++);
recordConsumer.startField("myarrayofoptional", index);
recordConsumer.startGroup();
recordConsumer.startField("list", 0);
for (Integer val : (Integer[]) record.get("myarrayofoptional")) {
recordConsumer.startGroup();
if (val != null) {
recordConsumer.startField("element", 0);
recordConsumer.addInteger(val);
recordConsumer.endField("element", 0);
}
recordConsumer.endGroup();
}
recordConsumer.endField("list", 0);
recordConsumer.endGroup();
recordConsumer.endField("myarrayofoptional", index++);
recordConsumer.startField("myrecordarray", index);
recordConsumer.startGroup();
recordConsumer.startField("array", 0);
recordConsumer.startGroup();
recordConsumer.startField("a", 0);
for (int val : (int[]) record.get("myrecordarraya")) {
recordConsumer.addInteger(val);
}
recordConsumer.endField("a", 0);
recordConsumer.startField("b", 1);
for (int val : (int[]) record.get("myrecordarrayb")) {
recordConsumer.addInteger(val);
}
recordConsumer.endField("b", 1);
recordConsumer.endGroup();
recordConsumer.endField("array", 0);
recordConsumer.endGroup();
recordConsumer.endField("myrecordarray", index++);
recordConsumer.startField("mymap", index);
recordConsumer.startGroup();
recordConsumer.startField("map", 0);
recordConsumer.startGroup();
Map<String, Integer> mymap = (Map<String, Integer>) record.get("mymap");
recordConsumer.startField("key", 0);
for (String key : mymap.keySet()) {
recordConsumer.addBinary(Binary.fromString(key));
}
recordConsumer.endField("key", 0);
recordConsumer.startField("value", 1);
for (int val : mymap.values()) {
recordConsumer.addInteger(val);
}
recordConsumer.endField("value", 1);
recordConsumer.endGroup();
recordConsumer.endField("map", 0);
recordConsumer.endGroup();
recordConsumer.endField("mymap", index++);
recordConsumer.startField("myfixed", index);
recordConsumer.addBinary(
Binary.fromReusedByteArray((byte[]) record.get("myfixed")));
recordConsumer.endField("myfixed", index++);
recordConsumer.endMessage();
}
});
Map<String, Object> record = new HashMap<String, Object>();
record.put("myboolean", true);
record.put("myint", 1);
record.put("mylong", 2L);
record.put("myfloat", 3.1f);
record.put("mydouble", 4.1);
record.put("mybytes", ByteBuffer.wrap("hello".getBytes(Charsets.UTF_8)));
record.put("mystring", "hello");
record.put("myenum", "a");
record.put("mynestedint", 1);
record.put("myarray", new int[] {1, 2, 3});
record.put("myoptionalarray", new int[] {1, 2, 3});
record.put("myarrayofoptional", new Integer[] {1, null, 2, null, 3});
record.put("myrecordarraya", new int[] {1, 2, 3});
record.put("myrecordarrayb", new int[] {4, 5, 6});
record.put("mymap", ImmutableMap.of("a", 1, "b", 2));
record.put("myfixed", new byte[] {(byte) 65});
parquetWriter.write(record);
parquetWriter.close();
Schema nestedRecordSchema = Schema.createRecord("mynestedrecord", null, null, false);
nestedRecordSchema.setFields(
Arrays.asList(new Schema.Field("mynestedint", Schema.create(Schema.Type.INT), null, null)));
GenericData.Record nestedRecord =
new GenericRecordBuilder(nestedRecordSchema).set("mynestedint", 1).build();
List<Integer> integerArray = Arrays.asList(1, 2, 3);
Schema recordArraySchema = Schema.createRecord("array", null, null, false);
recordArraySchema.setFields(
Arrays.asList(
new Schema.Field("a", Schema.create(Schema.Type.INT), null, null),
new Schema.Field("b", Schema.create(Schema.Type.INT), null, null)));
GenericRecordBuilder builder = new GenericRecordBuilder(recordArraySchema);
List<GenericData.Record> recordArray = new ArrayList<GenericData.Record>();
recordArray.add(builder.set("a", 1).set("b", 4).build());
recordArray.add(builder.set("a", 2).set("b", 5).build());
recordArray.add(builder.set("a", 3).set("b", 6).build());
GenericData.Array<GenericData.Record> genericRecordArray =
new GenericData.Array<GenericData.Record>(
Schema.createArray(recordArraySchema), recordArray);
GenericFixed genericFixed =
new GenericData.Fixed(Schema.createFixed("fixed", null, null, 1), new byte[] {(byte) 65});
// 3-level lists are deserialized with the extra layer present
Schema elementSchema = record("list", optionalField("element", primitive(Schema.Type.INT)));
GenericRecordBuilder elementBuilder = new GenericRecordBuilder(elementSchema);
GenericData.Array<GenericData.Record> genericRecordArrayWithNullIntegers =
new GenericData.Array<GenericData.Record>(
array(elementSchema),
Arrays.asList(
elementBuilder.set("element", 1).build(),
elementBuilder.set("element", null).build(),
elementBuilder.set("element", 2).build(),
elementBuilder.set("element", null).build(),
elementBuilder.set("element", 3).build()));
AvroParquetReader<GenericRecord> reader = new AvroParquetReader<GenericRecord>(testConf, file);
GenericRecord nextRecord = reader.read();
assertNotNull(nextRecord);
assertEquals(true, nextRecord.get("myboolean"));
assertEquals(1, nextRecord.get("myint"));
assertEquals(2L, nextRecord.get("mylong"));
assertEquals(3.1f, nextRecord.get("myfloat"));
assertEquals(4.1, nextRecord.get("mydouble"));
assertEquals(ByteBuffer.wrap("hello".getBytes(Charsets.UTF_8)), nextRecord.get("mybytes"));
assertEquals(str("hello"), nextRecord.get("mystring"));
assertEquals(str("a"), nextRecord.get("myenum"));
assertEquals(nestedRecord, nextRecord.get("mynestedrecord"));
assertEquals(integerArray, nextRecord.get("myarray"));
assertEquals(integerArray, nextRecord.get("myoptionalarray"));
assertEquals(genericRecordArrayWithNullIntegers, nextRecord.get("myarrayofoptional"));
assertEquals(genericRecordArray, nextRecord.get("myrecordarray"));
assertEquals(ImmutableMap.of(str("a"), 1, str("b"), 2), nextRecord.get("mymap"));
assertEquals(genericFixed, nextRecord.get("myfixed"));
}
@Test
public void testArrayWithNullValues() throws Exception {
Schema schema = new Schema.Parser().parse(Resources.getResource("all.avsc").openStream());
File tmp = File.createTempFile(getClass().getSimpleName(), ".tmp");
tmp.deleteOnExit();
tmp.delete();
Path file = new Path(tmp.getPath());
GenericData.Record nestedRecord =
new GenericRecordBuilder(schema.getField("mynestedrecord").schema())
.set("mynestedint", 1)
.build();
List<Integer> integerArray = Arrays.asList(1, 2, 3);
GenericData.Array<Integer> genericIntegerArray =
new GenericData.Array<Integer>(
Schema.createArray(Schema.create(Schema.Type.INT)), integerArray);
GenericFixed genericFixed =
new GenericData.Fixed(Schema.createFixed("fixed", null, null, 1), new byte[] {(byte) 65});
List<Integer> emptyArray = new ArrayList<Integer>();
ImmutableMap emptyMap = new ImmutableMap.Builder<String, Integer>().build();
Schema arrayOfOptionalIntegers = Schema.createArray(optional(Schema.create(Schema.Type.INT)));
GenericData.Array<Integer> genericIntegerArrayWithNulls =
new GenericData.Array<Integer>(arrayOfOptionalIntegers, Arrays.asList(1, null, 2, null, 3));
GenericData.Record record =
new GenericRecordBuilder(schema)
.set("mynull", null)
.set("myboolean", true)
.set("myint", 1)
.set("mylong", 2L)
.set("myfloat", 3.1f)
.set("mydouble", 4.1)
.set("mybytes", ByteBuffer.wrap("hello".getBytes(Charsets.UTF_8)))
.set("mystring", "hello")
.set("mynestedrecord", nestedRecord)
.set("myenum", "a")
.set("myarray", genericIntegerArray)
.set("myemptyarray", emptyArray)
.set("myoptionalarray", genericIntegerArray)
.set("myarrayofoptional", genericIntegerArrayWithNulls)
.set("mymap", ImmutableMap.of("a", 1, "b", 2))
.set("myemptymap", emptyMap)
.set("myfixed", genericFixed)
.build();
final AvroParquetWriter<GenericRecord> writer =
new AvroParquetWriter<GenericRecord>(file, schema);
try {
writer.write(record);
fail("Should not succeed writing an array with null values");
} catch (Exception e) {
Assert.assertTrue(
"Error message should provide context and help",
e.getMessage().contains("parquet.avro.write-old-list-structure"));
} finally {
writer.close();
}
}
@Test
public void testAll() throws Exception {
Schema schema = new Schema.Parser().parse(Resources.getResource("all.avsc").openStream());
File tmp = File.createTempFile(getClass().getSimpleName(), ".tmp");
tmp.deleteOnExit();
tmp.delete();
Path file = new Path(tmp.getPath());
AvroParquetWriter<GenericRecord> writer = new AvroParquetWriter<GenericRecord>(file, schema);
GenericData.Record nestedRecord =
new GenericRecordBuilder(schema.getField("mynestedrecord").schema())
.set("mynestedint", 1)
.build();
List<Integer> integerArray = Arrays.asList(1, 2, 3);
GenericData.Array<Integer> genericIntegerArray =
new GenericData.Array<Integer>(
Schema.createArray(Schema.create(Schema.Type.INT)), integerArray);
GenericFixed genericFixed =
new GenericData.Fixed(Schema.createFixed("fixed", null, null, 1), new byte[] {(byte) 65});
List<Integer> emptyArray = new ArrayList<Integer>();
ImmutableMap emptyMap = new ImmutableMap.Builder<String, Integer>().build();
GenericData.Record record =
new GenericRecordBuilder(schema)
.set("mynull", null)
.set("myboolean", true)
.set("myint", 1)
.set("mylong", 2L)
.set("myfloat", 3.1f)
.set("mydouble", 4.1)
.set("mybytes", ByteBuffer.wrap("hello".getBytes(Charsets.UTF_8)))
.set("mystring", "hello")
.set("mynestedrecord", nestedRecord)
.set("myenum", "a")
.set("myarray", genericIntegerArray)
.set("myemptyarray", emptyArray)
.set("myoptionalarray", genericIntegerArray)
.set("myarrayofoptional", genericIntegerArray)
.set("mymap", ImmutableMap.of("a", 1, "b", 2))
.set("myemptymap", emptyMap)
.set("myfixed", genericFixed)
.build();
writer.write(record);
writer.close();
AvroParquetReader<GenericRecord> reader = new AvroParquetReader<GenericRecord>(testConf, file);
GenericRecord nextRecord = reader.read();
Object expectedEnumSymbol =
compat ? "a" : new GenericData.EnumSymbol(schema.getField("myenum").schema(), "a");
assertNotNull(nextRecord);
assertEquals(null, nextRecord.get("mynull"));
assertEquals(true, nextRecord.get("myboolean"));
assertEquals(1, nextRecord.get("myint"));
assertEquals(2L, nextRecord.get("mylong"));
assertEquals(3.1f, nextRecord.get("myfloat"));
assertEquals(4.1, nextRecord.get("mydouble"));
assertEquals(ByteBuffer.wrap("hello".getBytes(Charsets.UTF_8)), nextRecord.get("mybytes"));
assertEquals(str("hello"), nextRecord.get("mystring"));
assertEquals(expectedEnumSymbol, nextRecord.get("myenum"));
assertEquals(nestedRecord, nextRecord.get("mynestedrecord"));
assertEquals(integerArray, nextRecord.get("myarray"));
assertEquals(emptyArray, nextRecord.get("myemptyarray"));
assertEquals(integerArray, nextRecord.get("myoptionalarray"));
assertEquals(integerArray, nextRecord.get("myarrayofoptional"));
assertEquals(ImmutableMap.of(str("a"), 1, str("b"), 2), nextRecord.get("mymap"));
assertEquals(emptyMap, nextRecord.get("myemptymap"));
assertEquals(genericFixed, nextRecord.get("myfixed"));
}
@Test
public void testKeyValueInput() throws ClassNotFoundException, IOException, InterruptedException {
// Create a test input file.
File inputFile = createInputFile();
// Configure the job input.
Job job = new Job();
FileInputFormat.setInputPaths(job, new Path(inputFile.getAbsolutePath()));
job.setInputFormatClass(CombineAvroKeyValueInputFormat.class);
AvroJob.setInputKeySchema(job, Schema.create(Schema.Type.INT));
AvroJob.setInputValueSchema(job, Schema.create(Schema.Type.STRING));
// Configure a mapper.
job.setMapperClass(IndexMapper.class);
job.setMapOutputKeyClass(Text.class);
job.setMapOutputValueClass(IntWritable.class);
// Configure a reducer.
job.setReducerClass(IndexReducer.class);
job.setOutputKeyClass(Text.class);
job.setOutputValueClass(AvroValue.class);
AvroJob.setOutputValueSchema(job, Schema.createArray(Schema.create(Schema.Type.INT)));
// Configure the output format.
job.setOutputFormatClass(AvroKeyValueOutputFormat.class);
Path outputPath = new Path(mTempDir.getRoot().getPath(), "out-index");
FileOutputFormat.setOutputPath(job, outputPath);
// Run the job.
assertTrue(job.waitForCompletion(true));
// Verify that the output Avro container file as the expected data.
File avroFile = new File(outputPath.toString(), "part-r-00000.avro");
DatumReader<GenericRecord> datumReader =
new SpecificDatumReader<GenericRecord>(
AvroKeyValue.getSchema(
Schema.create(Schema.Type.STRING),
Schema.createArray(Schema.create(Schema.Type.INT))));
DataFileReader<GenericRecord> avroFileReader =
new DataFileReader<GenericRecord>(avroFile, datumReader);
assertTrue(avroFileReader.hasNext());
AvroKeyValue<CharSequence, List<Integer>> appleRecord =
new AvroKeyValue<CharSequence, List<Integer>>(avroFileReader.next());
assertNotNull(appleRecord.get());
assertEquals("apple", appleRecord.getKey().toString());
List<Integer> appleDocs = appleRecord.getValue();
assertEquals(3, appleDocs.size());
assertTrue(appleDocs.contains(1));
assertTrue(appleDocs.contains(2));
assertTrue(appleDocs.contains(3));
assertTrue(avroFileReader.hasNext());
AvroKeyValue<CharSequence, List<Integer>> bananaRecord =
new AvroKeyValue<CharSequence, List<Integer>>(avroFileReader.next());
assertNotNull(bananaRecord.get());
assertEquals("banana", bananaRecord.getKey().toString());
List<Integer> bananaDocs = bananaRecord.getValue();
assertEquals(2, bananaDocs.size());
assertTrue(bananaDocs.contains(1));
assertTrue(bananaDocs.contains(2));
assertTrue(avroFileReader.hasNext());
AvroKeyValue<CharSequence, List<Integer>> carrotRecord =
new AvroKeyValue<CharSequence, List<Integer>>(avroFileReader.next());
assertEquals("carrot", carrotRecord.getKey().toString());
List<Integer> carrotDocs = carrotRecord.getValue();
assertEquals(1, carrotDocs.size());
assertTrue(carrotDocs.contains(1));
assertFalse(avroFileReader.hasNext());
avroFileReader.close();
}
@Test
public void testConvertBigQuerySchemaToAvroSchema() {
TableSchema tableSchema = new TableSchema();
tableSchema.setFields(fields);
Schema avroSchema =
BigQueryAvroUtils.toGenericAvroSchema("testSchema", tableSchema.getFields());
assertThat(avroSchema.getField("number").schema(), equalTo(Schema.create(Type.LONG)));
assertThat(
avroSchema.getField("species").schema(),
equalTo(Schema.createUnion(Schema.create(Type.NULL), Schema.create(Type.STRING))));
assertThat(
avroSchema.getField("quality").schema(),
equalTo(Schema.createUnion(Schema.create(Type.NULL), Schema.create(Type.DOUBLE))));
assertThat(
avroSchema.getField("quantity").schema(),
equalTo(Schema.createUnion(Schema.create(Type.NULL), Schema.create(Type.LONG))));
assertThat(
avroSchema.getField("birthday").schema(),
equalTo(Schema.createUnion(Schema.create(Type.NULL), Schema.create(Type.LONG))));
assertThat(
avroSchema.getField("flighted").schema(),
equalTo(Schema.createUnion(Schema.create(Type.NULL), Schema.create(Type.BOOLEAN))));
assertThat(
avroSchema.getField("sound").schema(),
equalTo(Schema.createUnion(Schema.create(Type.NULL), Schema.create(Type.BYTES))));
assertThat(
avroSchema.getField("anniversaryDate").schema(),
equalTo(Schema.createUnion(Schema.create(Type.NULL), Schema.create(Type.STRING))));
assertThat(
avroSchema.getField("anniversaryDatetime").schema(),
equalTo(Schema.createUnion(Schema.create(Type.NULL), Schema.create(Type.STRING))));
assertThat(
avroSchema.getField("anniversaryTime").schema(),
equalTo(Schema.createUnion(Schema.create(Type.NULL), Schema.create(Type.STRING))));
assertThat(
avroSchema.getField("scion").schema(),
equalTo(
Schema.createUnion(
Schema.create(Type.NULL),
Schema.createRecord(
"scion",
"org.apache.beam.sdk.io.gcp.bigquery",
"Translated Avro Schema for scion",
false,
ImmutableList.of(
new Field(
"species",
Schema.createUnion(
Schema.create(Type.NULL), Schema.create(Type.STRING)),
null,
(Object) null))))));
assertThat(
avroSchema.getField("associates").schema(),
equalTo(
Schema.createArray(
Schema.createRecord(
"associates",
"org.apache.beam.sdk.io.gcp.bigquery",
"Translated Avro Schema for associates",
false,
ImmutableList.of(
new Field(
"species",
Schema.createUnion(
Schema.create(Type.NULL), Schema.create(Type.STRING)),
null,
(Object) null))))));
}
Schema computeAvroSchema() {
return Schema.createArray(bodyType.getAvroSchema());
}
