Tutorial: Building with CleaveDB

Group & Aggregate Stages

In addition to standard filtering and projection, CleaveDB allows powerful group-and-aggregate stages inside a pipeline. The output of an aggregation stage is itself a stream of structured records that can be further sorted, limited, or transformed by downstream THEN clauses.

Multi-metric group aggregation

Group documents by an attribute and compute multiple statistical metrics in one pass:

PIPE FROM emp
  THEN GROUP BY dept
    TALLY AS n,
    MIN OF age AS youngest,
    MAX OF age AS oldest,
    AVERAGE OF age AS avg

This groups all employee records by department and emits structured summary objects containing the count, minimum age, maximum age, and average age per department.

Sorting by computed aggregate aliases

Because stages feed forward into the next stage, you can sort and limit by the computed aliases produced during aggregation:

PIPE FROM orders
  THEN WHERE status = "completed"
  THEN GROUP BY region TOTAL OF revenue AS region_total
  THEN ARRANGED BY region_total GOING DOWN
  THEN LIMIT 5

Notice how ARRANGED BY region_total GOING DOWN directly targets the alias created by the preceding TOTAL OF revenue AS region_total stage, and THEN LIMIT 5 yields the top 5 highest-revenue regions.

Supported aggregate functions in PIPE