SELECT (Stream)

Continuously pulls records from the stream(s) specified. It is usually used in an interactive CLI to monitor realtime changes of data. Note that the query writes records to a random-named stream which is hidden from users.

Synopsis

  1. SELECT <* | expression [ AS field_alias ] [, ...]>
  2.   FROM stream_name [, ...]
  3.   [ WHERE search_condition ]
  4.   [ GROUP BY field_name [, window_type] ]
  5.   EMIT CHANGES;

Notes

  • expression can be a field name, a constant, or their association, such as temperature, weather.humidity, 114514, 1 + 2, SUM(productions) and `COUNT(*)` . Note that the last one is a raw column name which is used when a column name contains function names. See Special Characters.
  • some_interval represents a period of time. See Intervals.

  • window_type specifies the type of time window:

    1. window_type ::= TUMBLING some_interval
    2.               | HOPPING  some_interval some_interval
    3.               | SLIDING  some_interval

  • search_condition is actually a boolean expression:

    1. search_condition ::= [NOT] predicate [ <AND | OR> predicate [, ...] ]
    2. predicate        ::= expression comp_op expression
    3. comp_op          ::= = | <> | > | < | >= | <=

Examples

  • A simple query:
  1. SELECT * FROM my_stream EMIT CHANGES;
  • Filtering rows:
  1. SELECT temperature, humidity FROM weather WHERE temperature > 10 AND humidity < 75 EMIT CHANGES;
  • Joining streams:
  1. SELECT stream1.temperature, stream2.humidity FROM stream1, stream2 WHERE stream1.humidity = stream2.humidity EMIT CHANGES;
  • Grouping records:
  1. SELECT COUNT(*) FROM weather GROUP BY cityId, TUMBLING (INTERVAL 10 SECOND) EMIT CHANGES;