forked from D-Net/dnet-hadoop
11 lines
609 B
SQL
11 lines
609 B
SQL
-- LaReferencia Distinct
|
|
DROP TABLE IF EXISTS ${usageStatsDB}.lareferencialogdistinct;
|
|
|
|
CREATE TABLE IF NOT EXISTS ${usageStatsDB}.lareferencialogdistinct(matomoid INT, source INT, id_visit STRING, country STRING, action STRING, url STRING,
|
|
entity_id STRING, source_item_type STRING, timestamp STRING, referrer_name STRING, agent STRING)
|
|
clustered by (source, id_visit, action, timestamp, entity_id)
|
|
into 100 buckets stored as orc tblproperties('transactional'='true');
|
|
|
|
INSERT INTO ${usageStatsDB}.lareferencialogdistinct
|
|
SELECT DISTINCT * FROM ${usageRawDataDB}.lareferencialog WHERE entity_id is not null;
|