/**
* Wrap an avro schema as a nullable union if needed. For instance, wrap schema "int" as ["null",
* "int"]
*/
public static Schema wrapAsUnion(Schema schema, boolean nullable) {
if (nullable) {
/* if schema is an acceptable union, then return itself */
if (schema.getType().equals(Schema.Type.UNION) && isAcceptableUnion(schema)) return schema;
else return Schema.createUnion(Arrays.asList(NullSchema, schema));
} else
/*do not wrap it if not */
return schema;
}
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();
}
}
Schema computeAvroSchema() {
HashSet<String> observedSchemas = new HashSet<String>();
List<Schema> fields = new ArrayList<Schema>();
for (InferredType it : unionTypes) {
Schema itS = it.getAvroSchema();
if (itS == null) {
continue;
}
String schemaDesc = itS.toString();
if (!observedSchemas.contains(schemaDesc)) {
observedSchemas.add(schemaDesc);
fields.add(it.getAvroSchema());
}
}
return Schema.createUnion(fields);
}
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 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))))));
}
public static Schema generateAvroSchema(FieldType field) {
return Schema.createUnion(Lists.newArrayList(Schema.create(NULL), getAvroSchema(field)));
}