// get default marshallers and validators
List<AbstractType> marshallers = getDefaultMarshallers(cfDef);
Map<ByteBuffer,AbstractType> validators = getValidatorMap(cfDef);
// add key
ResourceFieldSchema keyFieldSchema = new ResourceFieldSchema();
keyFieldSchema.setName("key");
keyFieldSchema.setType(getPigType(marshallers.get(2)));
ResourceSchema bagSchema = new ResourceSchema();
ResourceFieldSchema bagField = new ResourceFieldSchema();
bagField.setType(DataType.BAG);
bagField.setName("columns");
// inside the bag, place one tuple with the default comparator/validator schema
ResourceSchema bagTupleSchema = new ResourceSchema();
ResourceFieldSchema bagTupleField = new ResourceFieldSchema();
bagTupleField.setType(DataType.TUPLE);
ResourceFieldSchema bagcolSchema = new ResourceFieldSchema();
ResourceFieldSchema bagvalSchema = new ResourceFieldSchema();
bagcolSchema.setName("name");
bagvalSchema.setName("value");
bagcolSchema.setType(getPigType(marshallers.get(0)));
bagvalSchema.setType(getPigType(marshallers.get(1)));
bagTupleSchema.setFields(new ResourceFieldSchema[] { bagcolSchema, bagvalSchema });
bagTupleField.setSchema(bagTupleSchema);
bagSchema.setFields(new ResourceFieldSchema[] { bagTupleField });
bagField.setSchema(bagSchema);
// will contain all fields for this schema
List<ResourceFieldSchema> allSchemaFields = new ArrayList<ResourceFieldSchema>();
// add the key first, then the indexed columns, and finally the bag
allSchemaFields.add(keyFieldSchema);
// defined validators/indexes
for (ColumnDef cdef : cfDef.column_metadata)
{
// make a new tuple for each col/val pair
ResourceSchema innerTupleSchema = new ResourceSchema();
ResourceFieldSchema innerTupleField = new ResourceFieldSchema();
innerTupleField.setType(DataType.TUPLE);
innerTupleField.setSchema(innerTupleSchema);
innerTupleField.setName(new String(cdef.getName()));
ResourceFieldSchema idxColSchema = new ResourceFieldSchema();
idxColSchema.setName("name");
idxColSchema.setType(getPigType(marshallers.get(0)));
ResourceFieldSchema valSchema = new ResourceFieldSchema();
AbstractType validator = validators.get(cdef.name);
if (validator == null)
validator = marshallers.get(1);
valSchema.setName("value");
valSchema.setType(getPigType(validator));
innerTupleSchema.setFields(new ResourceFieldSchema[] { idxColSchema, valSchema });
allSchemaFields.add(innerTupleField);
}
// bag at the end for unknown columns
allSchemaFields.add(bagField);
// add top-level index elements if needed
if (usePartitionFilter)
{
for (ColumnDef cdef : getIndexes())
{
ResourceFieldSchema idxSchema = new ResourceFieldSchema();
idxSchema.setName("index_" + new String(cdef.getName()));
AbstractType validator = validators.get(cdef.name);
if (validator == null)
validator = marshallers.get(1);
idxSchema.setType(getPigType(validator));
allSchemaFields.add(idxSchema);
}
}
// top level schema contains everything
schema.setFields(allSchemaFields.toArray(new ResourceFieldSchema[allSchemaFields.size()]));