in src/main/java/org/apache/datasketches/pig/sampling/VarOptSampling.java [125:156]
public Schema outputSchema(final Schema input) {
try {
if ((input == null) || (input.size() == 0)) {
throw new IllegalArgumentException("Degenerate input schema to VarOptSampling");
}
// first element must be a bag, weightIdx_ element of tuples must be a float or double
if (input.getField(0).type != DataType.BAG) {
throw new IllegalArgumentException("VarOpt input must be a data bag: "
+ input.toString());
}
final Schema record = input.getField(0).schema; // record has a tuple in field 0
final Schema fields = record.getField(0).schema;
if ((fields.getField(this.weightIdx_).type != DataType.DOUBLE)
&& (fields.getField(this.weightIdx_).type != DataType.FLOAT)) {
throw new IllegalArgumentException("weightIndex item of VarOpt tuple must be a "
+ "weight (double/float), found " + fields.getField(0).type
+ ": " + fields.toString());
}
final Schema weightedSampleSchema = new Schema();
weightedSampleSchema.add(new Schema.FieldSchema(WEIGHT_ALIAS, DataType.DOUBLE));
weightedSampleSchema.add(new Schema.FieldSchema(RECORD_ALIAS, record, DataType.TUPLE));
return new Schema(new Schema.FieldSchema(getSchemaName(this
.getClass().getName().toLowerCase(), record), weightedSampleSchema, DataType.BAG));
}
catch (final FrontendException e) {
throw new RuntimeException(e);
}
}