All map entries must be of the same type. Regardless of the type of the map's values, its keys are always strings.

The string "[]" is reserved and must not be used for key names.

When you declare a table field as a map, you use the MAP() statement. You must also declare the map element's data types.

To define a simple two-field table where the primary key is a UID and the second field contains a map of integers, you use the following DDL statement:

CREATE TABLE myTable (
    uid INTEGER,
    myMap MAP(INTEGER),
    PRIMARY KEY (uid)
) 

CHECK constraints are supported for map fields. See CHECK for more information.

DEFAULT and NOT NULL constraints are not supported for map fields.

To write the map:

    mmap = {"field1" : 1, 
            "field2" : 2, 
            "field3" : 3}
     
    row_d = {'uid' : 0,
             'myMap' : mmap
            }
    try:
        store.put("myTable", row_d)
        logging.debug("Store write succeeded.")
    except IllegalArgumentException, iae:
        logging.error("Could not write table.")
        logging.error(iae.message)
        sys.exit(-1) 

To read map field2:

    try:
        primary_key_d = {"uid" : 0}
        row = store.get("myTable", primary_key_d)
        if not row:
            logging.debug("Row retrieval failed")
        else:
            logging.debug("Row retrieval succeeded.")
            ## prints '2'
            print row['myMap']['field2']
    except IllegalArgumentException, iae:
        logging.error("Row retrieval failed.")
        logging.error(iae.message)
        return
    except KeyError, ke:
        logging.error("Row display failed. Bad key: %s" % ke.message)