Stream specs

Data types

In eKuiper, each column or an expression has a related data type. A data type describes (and constrains) the set of values that a column of that type can hold or an expression of that type can produce.

Below is the list of data types supported.

Language definitions

CREATE STREAM   
    stream_name   
    ( column_name <data_type> [ ,...n ] )
    WITH ( property_name = expression [, ...] );

The supported property names.

Example 1,

my_stream 
  (id bigint, name string, score float)
WITH ( datasource = "topic/temperature", FORMAT = "json", KEY = "id");

The stream will subscribe to MQTT topic topic/temperature, the server connection uses server key of default section in configuration file $ekuiper/etc/mqtt_source.yaml.

• See MQTT source for more info.

Example 2,

demo (
        USERID BIGINT,
        FIRST_NAME STRING,
        LAST_NAME STRING,
        NICKNAMES ARRAY(STRING),
        Gender BOOLEAN,
        ADDRESS STRUCT(STREET_NAME STRING, NUMBER BIGINT),
    ) WITH (DATASOURCE="test/", FORMAT="JSON", KEY="USERID", CONF_KEY="demo");

The stream will subscribe to MQTT topic test/, the server connection uses settings of demo section in configuration file $ekuiper/etc/mqtt_source.yaml.

• See MQTT source for more info.

• See rules and streams CLI docs for more information of rules & streams management.

Share source instance across rules

By default, each rule will instantiate its own source instance. In some scenarios, users may need to manipulate the exact same data stream with different rules. For example, for the data of temperature from a sensor. They may want to trigger an alert when the average for a period of time is higher than 30 degree and trigger another alert when it is lower than 0. With default configuration, each rule creates a source instance and may receive data in different order due to network delay or other factors so that the average calculation may happen with different context. By sharing the instance, we can assure both rules are processing the same data. Additionally, it will have better performance by eliminating the overhead of instantiation.

To use the share instance mode, just set the SHARED option to true in the stream definition.

demo (
        ...
    ) WITH (DATASOURCE="test", FORMAT="JSON", KEY="USERID", SHARED="true");

Strict Validation

The value of StrictValidation can be true or false.
1) True: Drop the message if the message  is not satisfy with the stream definition.
2) False: Keep the message, but fill the missing field with default empty value.
bigint: 0
float: 0.0
string: ""
datetime: the current time
boolean: false
bytea: nil
array: zero length array
struct: null value

Schema-less stream

If the data type of the stream is unknown or varying, we can define it without the fields. This is called schema-less. It is defined by leaving the fields empty.

schemaless_stream 
  ()
WITH ( datasource = "topic/temperature", FORMAT = "json", KEY = "id");

Schema-less stream field data type will be determined at runtime. If the field is used in an incompatible clause, a runtime error will be thrown and send to the sink. For example, where temperature > 30. Once a temperature is not a number, an error will be sent to the sink.

See Query languange element for more inforamtion of SQL language.

Binary Stream

Specify “BINARY” format for streams of binary data such as image or video streams. The payload of such streams is a block of binary data without fields. So it is required to define the stream as only one field of bytea. In the below example, the payload will be parsed into image field of demoBin stream.

demoBin (
    image BYTEA
) WITH (DATASOURCE="test/", FORMAT="BINARY");

If “BINARY” format stream is defined as schemaless, a default field named self will be assigned for the binary payload.