@Override
public PCollection<T> apply(PInput input) {
if (filepattern == null) {
throw new IllegalStateException(
"need to set the filepattern of an AvroIO.Read transform");
}
if (schema == null) {
throw new IllegalStateException("need to set the schema of an AvroIO.Read transform");
}
if (validate) {
try {
checkState(
!IOChannelUtils.getFactory(filepattern).match(filepattern).isEmpty(),
"Unable to find any files matching %s",
filepattern);
} catch (IOException e) {
throw new IllegalStateException(String.format("Failed to validate %s", filepattern), e);
}
}
@SuppressWarnings("unchecked")
Bounded<T> read =
type == GenericRecord.class
? (Bounded<T>)
com.google.cloud.dataflow.sdk.io.Read.from(
AvroSource.from(filepattern).withSchema(schema))
: com.google.cloud.dataflow.sdk.io.Read.from(
AvroSource.from(filepattern).withSchema(type));
PCollection<T> pcol = input.getPipeline().apply("Read", read);
// Honor the default output coder that would have been used by this PTransform.
pcol.setCoder(getDefaultOutputCoder());
return pcol;
}
@Override
public SinkWriter<WindowedValue<KV<K, V>>> writer() throws IOException {
return new IsmSinkWriter(IOChannelUtils.create(filename, MimeTypes.BINARY));
}