HPL/SQL - Procedural SQL on Hadoop, NoSQL and RDBMS

====== PART_COUNT_BY Function ====== PART_COUNT_BY function returns the number partitions grouped by specified partition columns in the table. **Syntax**: <code language="sql"> PART_COUNT_BY([db_name.]table_name, [part_col, ...]); </code> **Parameters:** | [dbname.]table_name | Identifier, variable or expression | | part_col | One or more partition columns used for aggregation | **Return Value:** * The number of top-level partitions if //part_col// is not specified * Partition value and total number of existing partitions with the same value if //part_col// is not specified **Examples:** Table //db.orders// is partitioned by //dt// and //region// columns and has the following partitions: | dt=2015-09-01/region=1 | | dt=2015-09-01/region=2 | | dt=2015-09-02/region=1 | | dt=2015-09-03/region=3 | | dt=2015-09-03/region=2 | Get the total number of distinct top-level partitions (dt=2015-09-01, dt=2015-09-02 and dt=2015-09-03): <code language="sql"> part_count_by(db.orders); -- 3 </code> Get the top level partitions and the number of its sub-partitions: <code language="sql"> part_count_by(db.orders, dt); -- dt=2015-09-01 2 dt=2015-09-02 1 dt=2015-09-03 2 </code> Get the number of specified partitions: <code language="sql"> part_count_by(db.orders, region); -- region=1 2 region=2 2 region=3 1 </code> **Compatibility**: HPL/SQL extension. **Version:** HPL/SQL 0.3.13 See also: * [[part-count|PART_COUNT]] - Get the number of partitions * [[part-loc|PART_LOC]] - Get the location of a partition