forked from D-Net/dnet-hadoop
fixed formatting
This commit is contained in:
parent
2c4dcc90ba
commit
1c029b9fc0
|
|
@ -3,14 +3,37 @@
|
|||
-- Tables/views from external tables/views (Fundref, Country, CountyGDP, roarmap, rndexpediture)
|
||||
------------------------------------------------------------------------------------------------
|
||||
------------------------------------------------------------------------------------------------
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.fundref AS SELECT * FROM ${external_stats_db_name}.fundref;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.country AS SELECT * FROM ${external_stats_db_name}.country;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.countrygdp AS SELECT * FROM ${external_stats_db_name}.countrygdp;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.roarmap AS SELECT * FROM ${external_stats_db_name}.roarmap;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.rndexpediture AS SELECT * FROM ${external_stats_db_name}.rndexpediture;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.context AS SELECT * FROM ${external_stats_db_name}.context;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.category AS SELECT * FROM ${external_stats_db_name}.category;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.concept AS SELECT * FROM ${external_stats_db_name}.concept;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.fundref AS
|
||||
SELECT *
|
||||
FROM ${external_stats_db_name}.fundref;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.country AS
|
||||
SELECT *
|
||||
FROM ${external_stats_db_name}.country;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.countrygdp AS
|
||||
SELECT *
|
||||
FROM ${external_stats_db_name}.countrygdp;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.roarmap AS
|
||||
SELECT *
|
||||
FROM ${external_stats_db_name}.roarmap;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.rndexpediture AS
|
||||
SELECT *
|
||||
FROM ${external_stats_db_name}.rndexpediture;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.context AS
|
||||
SELECT *
|
||||
FROM ${external_stats_db_name}.context;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.category AS
|
||||
SELECT *
|
||||
FROM ${external_stats_db_name}.category;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.concept AS
|
||||
SELECT *
|
||||
FROM ${external_stats_db_name}.concept;
|
||||
|
||||
|
||||
------------------------------------------------------------------------------------------------
|
||||
|
|
@ -18,7 +41,8 @@ CREATE OR REPLACE VIEW ${stats_db_name}.concept AS SELECT * FROM ${external_sta
|
|||
-- Creation date of the database
|
||||
------------------------------------------------------------------------------------------------
|
||||
------------------------------------------------------------------------------------------------
|
||||
create table ${stats_db_name}.creation_date as select date_format(current_date(), 'dd-MM-yyyy') as date;
|
||||
create table ${stats_db_name}.creation_date as
|
||||
select date_format(current_date(), 'dd-MM-yyyy') as date;
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.creation_date COMPUTE STATISTICS;
|
||||
ANALYZE TABLE ${stats_db_name}.creation_date COMPUTE STATISTICS FOR COLUMNS;
|
||||
|
|
@ -5,43 +5,114 @@
|
|||
----------------------------------------------------------------
|
||||
|
||||
--Datasource temporary table updates
|
||||
UPDATE ${stats_db_name}.datasource_tmp SET harvested='true' WHERE datasource_tmp.id IN (SELECT DISTINCT d.id FROM ${stats_db_name}.datasource_tmp d, ${stats_db_name}.result_datasources rd WHERE d.id=rd.datasource);
|
||||
UPDATE ${stats_db_name}.datasource_tmp
|
||||
SET harvested='true'
|
||||
WHERE datasource_tmp.id IN (SELECT DISTINCT d.id
|
||||
FROM ${stats_db_name}.datasource_tmp d,
|
||||
${stats_db_name}.result_datasources rd
|
||||
WHERE d.id = rd.datasource);
|
||||
|
||||
-- Project temporary table update and final project table creation with final updates that can not be applied to ORC tables
|
||||
UPDATE ${stats_db_name}.project_tmp SET haspubs='yes' WHERE project_tmp.id IN (SELECT pr.id FROM ${stats_db_name}.project_results pr, ${stats_db_name}.result r WHERE pr.result=r.id AND r.type='publication');
|
||||
UPDATE ${stats_db_name}.project_tmp
|
||||
SET haspubs='yes'
|
||||
WHERE project_tmp.id IN (SELECT pr.id
|
||||
FROM ${stats_db_name}.project_results pr,
|
||||
${stats_db_name}.result r
|
||||
WHERE pr.result = r.id
|
||||
AND r.type = 'publication');
|
||||
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.project;
|
||||
CREATE TABLE ${stats_db_name}.project stored as parquet as
|
||||
SELECT p.id , p.acronym, p.title, p.funder, p.funding_lvl0, p.funding_lvl1, p.funding_lvl2, p.ec39, p.type, p.startdate, p.enddate, p.start_year, p.end_year, p.duration,
|
||||
CASE WHEN prr1.id IS NULL THEN 'no' ELSE 'yes' END AS haspubs,
|
||||
CASE WHEN prr1.id IS NULL THEN 0 ELSE prr1.np END AS numpubs,
|
||||
CASE WHEN prr2.id IS NULL THEN 0 ELSE prr2.daysForlastPub END AS daysforlastpub,
|
||||
CASE WHEN prr2.id IS NULL THEN 0 ELSE prr2.dp END AS delayedpubs,
|
||||
p.callidentifier, p.code
|
||||
SELECT p.id,
|
||||
p.acronym,
|
||||
p.title,
|
||||
p.funder,
|
||||
p.funding_lvl0,
|
||||
p.funding_lvl1,
|
||||
p.funding_lvl2,
|
||||
p.ec39,
|
||||
p.type,
|
||||
p.startdate,
|
||||
p.enddate,
|
||||
p.start_year,
|
||||
p.end_year,
|
||||
p.duration,
|
||||
CASE WHEN prr1.id IS NULL THEN 'no' ELSE 'yes' END AS haspubs,
|
||||
CASE WHEN prr1.id IS NULL THEN 0 ELSE prr1.np END AS numpubs,
|
||||
CASE WHEN prr2.id IS NULL THEN 0 ELSE prr2.daysForlastPub END AS daysforlastpub,
|
||||
CASE WHEN prr2.id IS NULL THEN 0 ELSE prr2.dp END AS delayedpubs,
|
||||
p.callidentifier,
|
||||
p.code
|
||||
FROM ${stats_db_name}.project_tmp p
|
||||
LEFT JOIN (SELECT pr.id, count(distinct pr.result) AS np
|
||||
FROM ${stats_db_name}.project_results pr INNER JOIN ${stats_db_name}.result r ON pr.result=r.id
|
||||
WHERE r.type='publication'
|
||||
LEFT JOIN (SELECT pr.id, count(distinct pr.result) AS np
|
||||
FROM ${stats_db_name}.project_results pr
|
||||
INNER JOIN ${stats_db_name}.result r ON pr.result = r.id
|
||||
WHERE r.type = 'publication'
|
||||
GROUP BY pr.id) AS prr1 on prr1.id = p.id
|
||||
LEFT JOIN (SELECT pp.id, max(datediff(to_date(r.date), to_date(pp.enddate)) ) AS daysForlastPub , count(distinct r.id) AS dp
|
||||
FROM ${stats_db_name}.project_tmp pp, ${stats_db_name}.project_results pr, ${stats_db_name}.result r
|
||||
WHERE pp.id=pr.id AND pr.result=r.id AND r.type='publication' AND datediff(to_date(r.date), to_date(pp.enddate)) > 0
|
||||
LEFT JOIN (SELECT pp.id,
|
||||
max(datediff(to_date(r.date), to_date(pp.enddate))) AS daysForlastPub,
|
||||
count(distinct r.id) AS dp
|
||||
FROM ${stats_db_name}.project_tmp pp,
|
||||
${stats_db_name}.project_results pr,
|
||||
${stats_db_name}.result r
|
||||
WHERE pp.id = pr.id
|
||||
AND pr.result = r.id
|
||||
AND r.type = 'publication'
|
||||
AND datediff(to_date(r.date), to_date(pp.enddate)) > 0
|
||||
GROUP BY pp.id) AS prr2
|
||||
ON prr2.id = p.id;
|
||||
|
||||
-- Publication temporary table updates
|
||||
UPDATE ${stats_db_name}.publication_tmp SET delayed = 'yes' WHERE publication_tmp.id IN (SELECT distinct r.id FROM stats_wf_db_obs.result r, ${stats_db_name}.project_results pr, ${stats_db_name}.project_tmp p WHERE r.id=pr.result AND pr.id=p.id AND to_date(r.date)-to_date(p.enddate) > 0);
|
||||
UPDATE ${stats_db_name}.publication_tmp
|
||||
SET delayed = 'yes'
|
||||
WHERE publication_tmp.id IN (SELECT distinct r.id
|
||||
FROM stats_wf_db_obs.result r,
|
||||
${stats_db_name}.project_results pr,
|
||||
${stats_db_name}.project_tmp p
|
||||
WHERE r.id = pr.result
|
||||
AND pr.id = p.id
|
||||
AND to_date(r.date) - to_date(p.enddate) > 0);
|
||||
|
||||
-- Dataset temporary table updates
|
||||
UPDATE ${stats_db_name}.dataset_tmp SET delayed = 'yes' WHERE dataset_tmp.id IN (SELECT distinct r.id FROM stats_wf_db_obs.result r, ${stats_db_name}.project_results pr, ${stats_db_name}.project_tmp p WHERE r.id=pr.result AND pr.id=p.id AND to_date(r.date)-to_date(p.enddate) > 0);
|
||||
UPDATE ${stats_db_name}.dataset_tmp
|
||||
SET delayed = 'yes'
|
||||
WHERE dataset_tmp.id IN (SELECT distinct r.id
|
||||
FROM stats_wf_db_obs.result r,
|
||||
${stats_db_name}.project_results pr,
|
||||
${stats_db_name}.project_tmp p
|
||||
WHERE r.id = pr.result
|
||||
AND pr.id = p.id
|
||||
AND to_date(r.date) - to_date(p.enddate) > 0);
|
||||
|
||||
-- Software temporary table updates
|
||||
UPDATE ${stats_db_name}.software_tmp SET delayed = 'yes' WHERE software_tmp.id IN (SELECT distinct r.id FROM ${stats_db_name}.result r, ${stats_db_name}.project_results pr, ${stats_db_name}.project_tmp p WHERE r.id=pr.result AND pr.id=p.id AND to_date(r.date)-to_date(p.enddate) > 0);
|
||||
UPDATE ${stats_db_name}.software_tmp
|
||||
SET delayed = 'yes'
|
||||
WHERE software_tmp.id IN (SELECT distinct r.id
|
||||
FROM ${stats_db_name}.result r,
|
||||
${stats_db_name}.project_results pr,
|
||||
${stats_db_name}.project_tmp p
|
||||
WHERE r.id = pr.result
|
||||
AND pr.id = p.id
|
||||
AND to_date(r.date) - to_date(p.enddate) > 0);
|
||||
|
||||
-- Oherresearchproduct temporary table updates
|
||||
UPDATE ${stats_db_name}.otherresearchproduct_tmp SET delayed = 'yes' WHERE otherresearchproduct_tmp.id IN (SELECT distinct r.id FROM ${stats_db_name}.result r, ${stats_db_name}.project_results pr, ${stats_db_name}.project_tmp p WHERE r.id=pr.result AND pr.id=p.id AND to_date(r.date)-to_date(p.enddate) > 0);
|
||||
UPDATE ${stats_db_name}.otherresearchproduct_tmp
|
||||
SET delayed = 'yes'
|
||||
WHERE otherresearchproduct_tmp.id IN (SELECT distinct r.id
|
||||
FROM ${stats_db_name}.result r,
|
||||
${stats_db_name}.project_results pr,
|
||||
${stats_db_name}.project_tmp p
|
||||
WHERE r.id = pr.result
|
||||
AND pr.id = p.id
|
||||
AND to_date(r.date) - to_date(p.enddate) > 0);
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.project_results_publication AS SELECT result_projects.id AS result, result_projects.project AS project_results, result.date as resultdate, project.enddate as projectenddate, result_projects.daysfromend AS daysfromend FROM ${stats_db_name}.result_projects, ${stats_db_name}.result, ${stats_db_name}.project WHERE result_projects.id=result.id AND result.type='publication' AND project.id=result_projects.project;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.project_results_publication AS
|
||||
SELECT result_projects.id AS result,
|
||||
result_projects.project AS project_results,
|
||||
result.date as resultdate,
|
||||
project.enddate as projectenddate,
|
||||
result_projects.daysfromend AS daysfromend
|
||||
FROM ${stats_db_name}.result_projects,
|
||||
${stats_db_name}.result,
|
||||
${stats_db_name}.project
|
||||
WHERE result_projects.id = result.id
|
||||
AND result.type = 'publication'
|
||||
AND project.id = result_projects.project;
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.project COMPUTE STATISTICS;
|
||||
ANALYZE TABLE ${stats_db_name}.project COMPUTE STATISTICS FOR COLUMNS;
|
||||
|
|
@ -1,21 +1,25 @@
|
|||
------------------------------------------------------------------------------------------------------
|
||||
-- Creating parquet tables from the updated temporary tables and removing unnecessary temporary tables
|
||||
------------------------------------------------------------------------------------------------------
|
||||
CREATE TABLE ${stats_db_name}.datasource stored AS parquet AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.datasource_tmp;
|
||||
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.datasource;
|
||||
CREATE TABLE ${stats_db_name}.datasource stored AS parquet AS SELECT * FROM ${stats_db_name}.datasource_tmp;
|
||||
CREATE TABLE ${stats_db_name}.publication stored AS parquet AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.publication_tmp;
|
||||
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.publication;
|
||||
CREATE TABLE ${stats_db_name}.publication stored AS parquet AS SELECT * FROM ${stats_db_name}.publication_tmp;
|
||||
CREATE TABLE ${stats_db_name}.dataset stored AS parquet AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.dataset_tmp;
|
||||
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.dataset;
|
||||
CREATE TABLE ${stats_db_name}.dataset stored AS parquet AS SELECT * FROM ${stats_db_name}.dataset_tmp;
|
||||
CREATE TABLE ${stats_db_name}.software stored AS parquet AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.software_tmp;
|
||||
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.software;
|
||||
CREATE TABLE ${stats_db_name}.software stored AS parquet AS SELECT * FROM ${stats_db_name}.software_tmp;
|
||||
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.otherresearchproduct;
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct stored AS parquet AS SELECT * FROM ${stats_db_name}.otherresearchproduct_tmp;
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct stored AS parquet AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.otherresearchproduct_tmp;
|
||||
|
||||
DROP TABLE ${stats_db_name}.project_tmp;
|
||||
DROP TABLE ${stats_db_name}.datasource_tmp;
|
||||
|
|
@ -29,13 +33,37 @@ DROP TABLE ${stats_db_name}.otherresearchproduct_tmp;
|
|||
---------------------------------------------
|
||||
|
||||
-- Result
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result AS SELECT *, bestlicence AS access_mode FROM ${stats_db_name}.publication UNION ALL SELECT *, bestlicence as access_mode FROM ${stats_db_name}.software UNION ALL SELECT *, bestlicence AS access_mode FROM ${stats_db_name}.dataset UNION ALL SELECT *, bestlicence AS access_mode FROM ${stats_db_name}.otherresearchproduct;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result AS
|
||||
SELECT *, bestlicence AS access_mode
|
||||
FROM ${stats_db_name}.publication
|
||||
UNION ALL
|
||||
SELECT *, bestlicence as access_mode
|
||||
FROM ${stats_db_name}.software
|
||||
UNION ALL
|
||||
SELECT *, bestlicence AS access_mode
|
||||
FROM ${stats_db_name}.dataset
|
||||
UNION ALL
|
||||
SELECT *, bestlicence AS access_mode
|
||||
FROM ${stats_db_name}.otherresearchproduct;
|
||||
|
||||
|
||||
-------------------------------------------------------------------------------
|
||||
-- To see with Antonis if the following is needed and where it should be placed
|
||||
-------------------------------------------------------------------------------
|
||||
CREATE TABLE ${stats_db_name}.numbers_country AS SELECT org.country AS country, count(distinct rd.datasource) AS datasources, count(distinct r.id) AS publications FROM ${stats_db_name}.result r, ${stats_db_name}.result_datasources rd, ${stats_db_name}.datasource d, ${stats_db_name}.datasource_organizations dor, ${stats_db_name}.organization org WHERE r.id=rd.id AND rd.datasource=d.id AND d.id=dor.id AND dor.organization=org.id AND r.type='publication' AND r.bestlicence='Open Access' GROUP BY org.country;
|
||||
CREATE TABLE ${stats_db_name}.numbers_country AS
|
||||
SELECT org.country AS country, count(distinct rd.datasource) AS datasources, count(distinct r.id) AS publications
|
||||
FROM ${stats_db_name}.result r,
|
||||
${stats_db_name}.result_datasources rd,
|
||||
${stats_db_name}.datasource d,
|
||||
${stats_db_name}.datasource_organizations dor,
|
||||
${stats_db_name}.organization org
|
||||
WHERE r.id = rd.id
|
||||
AND rd.datasource = d.id
|
||||
AND d.id = dor.id
|
||||
AND dor.organization = org.id
|
||||
AND r.type = 'publication'
|
||||
AND r.bestlicence = 'Open Access'
|
||||
GROUP BY org.country;
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.datasource COMPUTE STATISTICS;
|
||||
ANALYZE TABLE ${stats_db_name}.datasource COMPUTE STATISTICS FOR COLUMNS;
|
||||
|
|
|
|||
|
|
@ -5,41 +5,93 @@
|
|||
--------------------------------------------------------------
|
||||
|
||||
-- Publication temporary table
|
||||
CREATE TABLE ${stats_db_name}.publication_tmp (id STRING, title STRING, publisher STRING, journal STRING, date STRING, year STRING, bestlicence STRING, embargo_end_date STRING, delayed BOOLEAN, authors INT, source STRING, abstract BOOLEAN, type STRING ) clustered by (id) into 100 buckets stored as orc tblproperties('transactional'='true');
|
||||
CREATE TABLE ${stats_db_name}.publication_tmp
|
||||
(
|
||||
id STRING,
|
||||
title STRING,
|
||||
publisher STRING,
|
||||
journal STRING,
|
||||
date STRING,
|
||||
year STRING,
|
||||
bestlicence STRING,
|
||||
embargo_end_date STRING,
|
||||
delayed BOOLEAN,
|
||||
authors INT,
|
||||
source STRING,
|
||||
abstract BOOLEAN,
|
||||
type STRING
|
||||
)
|
||||
clustered by (id) into 100 buckets stored as orc tblproperties ('transactional' = 'true');
|
||||
|
||||
INSERT INTO ${stats_db_name}.publication_tmp SELECT substr(p.id, 4) as id, p.title[0].value as title, p.publisher.value as publisher, p.journal.name as journal ,
|
||||
p.dateofacceptance.value as date, date_format(p.dateofacceptance.value,'yyyy') as year, p.bestaccessright.classname as bestlicence,
|
||||
p.embargoenddate.value as embargo_end_date, false as delayed, size(p.author) as authors , concat_ws('\u003B',p.source.value) as source,
|
||||
case when size(p.description) > 0 then true else false end as abstract,
|
||||
'publication' as type
|
||||
INSERT INTO ${stats_db_name}.publication_tmp
|
||||
SELECT substr(p.id, 4) as id,
|
||||
p.title[0].value as title,
|
||||
p.publisher.value as publisher,
|
||||
p.journal.name as journal,
|
||||
p.dateofacceptance.value as date,
|
||||
date_format(p.dateofacceptance.value, 'yyyy') as year,
|
||||
p.bestaccessright.classname as bestlicence,
|
||||
p.embargoenddate.value as embargo_end_date,
|
||||
false as delayed,
|
||||
size(p.author) as authors,
|
||||
concat_ws('\u003B', p.source.value) as source,
|
||||
case when size(p.description) > 0 then true else false end as abstract,
|
||||
'publication' as type
|
||||
from ${openaire_db_name}.publication p
|
||||
where p.datainfo.deletedbyinference=false;
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.publication_classifications AS SELECT substr(p.id, 4) as id, instancetype.classname as type from ${openaire_db_name}.publication p LATERAL VIEW explode(p.instance.instancetype) instances as instancetype where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.publication_classifications AS
|
||||
SELECT substr(p.id, 4) as id, instancetype.classname as type
|
||||
from ${openaire_db_name}.publication p
|
||||
LATERAL VIEW explode(p.instance.instancetype) instances as instancetype
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.publication_concepts AS SELECT substr(p.id, 4) as id, contexts.context.id as concept from ${openaire_db_name}.publication p LATERAL VIEW explode(p.context) contexts as context where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.publication_concepts AS
|
||||
SELECT substr(p.id, 4) as id, contexts.context.id as concept
|
||||
from ${openaire_db_name}.publication p
|
||||
LATERAL VIEW explode(p.context) contexts as context
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.publication_datasources as
|
||||
SELECT p.id, case when d.id is null then 'other' else p.datasource end as datasource
|
||||
FROM (
|
||||
FROM (
|
||||
SELECT substr(p.id, 4) as id, substr(instances.instance.hostedby.key, 4) as datasource
|
||||
from ${openaire_db_name}.publication p lateral view explode(p.instance) instances as instance
|
||||
where p.datainfo.deletedbyinference=false ) p
|
||||
where p.datainfo.deletedbyinference = false) p
|
||||
LEFT OUTER JOIN (
|
||||
SELECT substr(d.id, 4) id
|
||||
from ${openaire_db_name}.datasource d
|
||||
WHERE d.datainfo.deletedbyinference=false ) d on p.datasource = d.id;
|
||||
WHERE d.datainfo.deletedbyinference = false) d on p.datasource = d.id;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.publication_languages AS select substr(p.id, 4) as id, p.language.classname as language FROM ${openaire_db_name}.publication p where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.publication_languages AS
|
||||
select substr(p.id, 4) as id, p.language.classname as language
|
||||
FROM ${openaire_db_name}.publication p
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.publication_oids AS SELECT substr(p.id, 4) AS id, oids.ids AS oid FROM ${openaire_db_name}.publication p LATERAL VIEW explode(p.originalid) oids AS ids where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.publication_oids AS
|
||||
SELECT substr(p.id, 4) AS id, oids.ids AS oid
|
||||
FROM ${openaire_db_name}.publication p
|
||||
LATERAL VIEW explode(p.originalid) oids AS ids
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.publication_pids AS SELECT substr(p.id, 4) AS id, ppid.qualifier.classname AS type, ppid.value as pid FROM ${openaire_db_name}.publication p LATERAL VIEW explode(p.pid) pids AS ppid where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.publication_pids AS
|
||||
SELECT substr(p.id, 4) AS id, ppid.qualifier.classname AS type, ppid.value as pid
|
||||
FROM ${openaire_db_name}.publication p
|
||||
LATERAL VIEW explode(p.pid) pids AS ppid
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.publication_topics as select substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS TYPE, subjects.subject.value AS topic FROM ${openaire_db_name}.publication p LATERAL VIEW explode(p.subject) subjects AS subject where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.publication_topics as
|
||||
select substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS TYPE, subjects.subject.value AS topic
|
||||
FROM ${openaire_db_name}.publication p
|
||||
LATERAL VIEW explode(p.subject) subjects AS subject
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
-- Publication_citations
|
||||
CREATE TABLE ${stats_db_name}.publication_citations AS SELECT substr(p.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS result FROM ${openaire_db_name}.publication p lateral view explode(p.extrainfo) citations AS citation WHERE xpath_string(citation.value, "//citation/id[@type='openaire']/@value") !="" and p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.publication_citations AS
|
||||
SELECT substr(p.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS result
|
||||
FROM ${openaire_db_name}.publication p
|
||||
lateral view explode(p.extrainfo) citations AS citation
|
||||
WHERE xpath_string(citation.value, "//citation/id[@type='openaire']/@value") != ""
|
||||
and p.datainfo.deletedbyinference = false;
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.publication_tmp COMPUTE STATISTICS;
|
||||
ANALYZE TABLE ${stats_db_name}.publication_tmp COMPUTE STATISTICS FOR COLUMNS;
|
||||
|
|
|
|||
|
|
@ -5,34 +5,94 @@
|
|||
------------------------------------------------------
|
||||
|
||||
-- Dataset temporary table supporting updates
|
||||
CREATE TABLE ${stats_db_name}.dataset_tmp (id STRING, title STRING, publisher STRING, journal STRING, date STRING, year STRING, bestlicence STRING, embargo_end_date STRING, delayed BOOLEAN, authors INT, source STRING, abstract BOOLEAN, type STRING ) clustered by (id) into 100 buckets stored AS orc tblproperties('transactional'='true');
|
||||
CREATE TABLE ${stats_db_name}.dataset_tmp
|
||||
(
|
||||
id STRING,
|
||||
title STRING,
|
||||
publisher STRING,
|
||||
journal STRING,
|
||||
date STRING,
|
||||
year STRING,
|
||||
bestlicence STRING,
|
||||
embargo_end_date STRING,
|
||||
delayed BOOLEAN,
|
||||
authors INT,
|
||||
source STRING,
|
||||
abstract BOOLEAN,
|
||||
type STRING
|
||||
)
|
||||
clustered by (id) into 100 buckets stored AS orc tblproperties ('transactional' = 'true');
|
||||
|
||||
INSERT INTO ${stats_db_name}.dataset_tmp SELECT substr(d.id, 4) AS id, d.title[0].value AS title, d.publisher.value AS publisher, cast(null AS string) AS journal,
|
||||
d.dateofacceptance.value as date, date_format(d.dateofacceptance.value,'yyyy') AS year, d.bestaccessright.classname AS bestlicence,
|
||||
d.embargoenddate.value AS embargo_end_date, false AS delayed, size(d.author) AS authors , concat_ws('\u003B',d.source.value) AS source,
|
||||
INSERT INTO ${stats_db_name}.dataset_tmp
|
||||
SELECT substr(d.id, 4) AS id,
|
||||
d.title[0].value AS title,
|
||||
d.publisher.value AS publisher,
|
||||
cast(null AS string) AS journal,
|
||||
d.dateofacceptance.value as date,
|
||||
date_format(d.dateofacceptance.value, 'yyyy') AS year,
|
||||
d.bestaccessright.classname AS bestlicence,
|
||||
d.embargoenddate.value AS embargo_end_date,
|
||||
false AS delayed,
|
||||
size(d.author) AS authors,
|
||||
concat_ws('\u003B', d.source.value) AS source,
|
||||
CASE WHEN SIZE(d.description) > 0 THEN TRUE ELSE FALSE end AS abstract,
|
||||
'dataset' AS type
|
||||
'dataset' AS type
|
||||
FROM ${openaire_db_name}.dataset d
|
||||
WHERE d.datainfo.deletedbyinference=FALSE;
|
||||
WHERE d.datainfo.deletedbyinference = FALSE;
|
||||
|
||||
-- Dataset_citations
|
||||
CREATE TABLE ${stats_db_name}.dataset_citations AS SELECT substr(d.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS result FROM ${openaire_db_name}.dataset d LATERAL VIEW explode(d.extrainfo) citations AS citation WHERE xpath_string(citation.value, "//citation/id[@type='openaire']/@value") !="" and d.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.dataset_citations AS
|
||||
SELECT substr(d.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS result
|
||||
FROM ${openaire_db_name}.dataset d
|
||||
LATERAL VIEW explode(d.extrainfo) citations AS citation
|
||||
WHERE xpath_string(citation.value, "//citation/id[@type='openaire']/@value") != ""
|
||||
and d.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.dataset_classifications AS SELECT substr(p.id, 4) AS id, instancetype.classname AS type FROM ${openaire_db_name}.dataset p LATERAL VIEW explode(p.instance.instancetype) instances AS instancetype where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.dataset_classifications AS
|
||||
SELECT substr(p.id, 4) AS id, instancetype.classname AS type
|
||||
FROM ${openaire_db_name}.dataset p
|
||||
LATERAL VIEW explode(p.instance.instancetype) instances AS instancetype
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.dataset_concepts AS SELECT substr(p.id, 4) as id, contexts.context.id as concept from ${openaire_db_name}.dataset p LATERAL VIEW explode(p.context) contexts as context where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.dataset_concepts AS
|
||||
SELECT substr(p.id, 4) as id, contexts.context.id as concept
|
||||
from ${openaire_db_name}.dataset p
|
||||
LATERAL VIEW explode(p.context) contexts as context
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.dataset_datasources AS SELECT p.id, case when d.id IS NULL THEN 'other' ELSE p.datasource END AS datasource FROM (SELECT substr(p.id, 4) as id, substr(instances.instance.hostedby.key, 4) AS datasource
|
||||
FROM ${openaire_db_name}.dataset p LATERAL VIEW explode(p.instance) instances AS instance where p.datainfo.deletedbyinference=false) p LEFT OUTER JOIN
|
||||
(SELECT substr(d.id, 4) id FROM ${openaire_db_name}.datasource d WHERE d.datainfo.deletedbyinference=false) d ON p.datasource = d.id;
|
||||
CREATE TABLE ${stats_db_name}.dataset_datasources AS
|
||||
SELECT p.id, case when d.id IS NULL THEN 'other' ELSE p.datasource END AS datasource
|
||||
FROM (
|
||||
SELECT substr(p.id, 4) as id, substr(instances.instance.hostedby.key, 4) AS datasource
|
||||
FROM ${openaire_db_name}.dataset p
|
||||
LATERAL VIEW explode(p.instance) instances AS instance
|
||||
where p.datainfo.deletedbyinference = false) p
|
||||
LEFT OUTER JOIN (
|
||||
SELECT substr(d.id, 4) id
|
||||
FROM ${openaire_db_name}.datasource d
|
||||
WHERE d.datainfo.deletedbyinference = false) d ON p.datasource = d.id;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.dataset_languages AS SELECT substr(p.id, 4) AS id, p.language.classname AS language FROM ${openaire_db_name}.dataset p where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.dataset_languages AS
|
||||
SELECT substr(p.id, 4) AS id, p.language.classname AS language
|
||||
FROM ${openaire_db_name}.dataset p
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.dataset_oids AS SELECT substr(p.id, 4) AS id, oids.ids AS oid FROM ${openaire_db_name}.dataset p LATERAL VIEW explode(p.originalid) oids AS ids where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.dataset_oids AS
|
||||
SELECT substr(p.id, 4) AS id, oids.ids AS oid
|
||||
FROM ${openaire_db_name}.dataset p
|
||||
LATERAL VIEW explode(p.originalid) oids AS ids
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.dataset_pids AS SELECT substr(p.id, 4) AS id, ppid.qualifier.classname AS type, ppid.value AS pid FROM ${openaire_db_name}.dataset p LATERAL VIEW explode(p.pid) pids AS ppid where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.dataset_pids AS
|
||||
SELECT substr(p.id, 4) AS id, ppid.qualifier.classname AS type, ppid.value AS pid
|
||||
FROM ${openaire_db_name}.dataset p
|
||||
LATERAL VIEW explode(p.pid) pids AS ppid
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.dataset_topics AS SELECT substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS type, subjects.subject.value AS topic FROM ${openaire_db_name}.dataset p LATERAL VIEW explode(p.subject) subjects AS subject where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.dataset_topics AS
|
||||
SELECT substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS type, subjects.subject.value AS topic
|
||||
FROM ${openaire_db_name}.dataset p
|
||||
LATERAL VIEW explode(p.subject) subjects AS subject
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.dataset_tmp COMPUTE STATISTICS;
|
||||
ANALYZE TABLE ${stats_db_name}.dataset_tmp COMPUTE STATISTICS FOR COLUMNS;
|
||||
|
|
|
|||
|
|
@ -5,34 +5,94 @@
|
|||
--------------------------------------------------------
|
||||
|
||||
-- Software temporary table supporting updates
|
||||
CREATE TABLE ${stats_db_name}.software_tmp (id STRING, title STRING, publisher STRING, journal STRING, date STRING, year STRING, bestlicence STRING, embargo_end_date STRING, delayed BOOLEAN, authors INT, source STRING, abstract BOOLEAN, type STRING ) clustered by (id) INTO 100 buckets stored AS orc tblproperties('transactional'='true');
|
||||
CREATE TABLE ${stats_db_name}.software_tmp
|
||||
(
|
||||
id STRING,
|
||||
title STRING,
|
||||
publisher STRING,
|
||||
journal STRING,
|
||||
date STRING,
|
||||
year STRING,
|
||||
bestlicence STRING,
|
||||
embargo_end_date STRING,
|
||||
delayed BOOLEAN,
|
||||
authors INT,
|
||||
source STRING,
|
||||
abstract BOOLEAN,
|
||||
type STRING
|
||||
)
|
||||
clustered by (id) INTO 100 buckets stored AS orc tblproperties ('transactional' = 'true');
|
||||
|
||||
INSERT INTO ${stats_db_name}.software_tmp SELECT substr(s.id, 4) as id, s.title[0].value AS title, s.publisher.value AS publisher, CAST(NULL AS string) AS journal,
|
||||
s.dateofacceptance.value AS DATE, date_format(s.dateofacceptance.value,'yyyy') AS YEAR, s.bestaccessright.classname AS bestlicence,
|
||||
s.embargoenddate.value AS embargo_end_date, FALSE AS delayed, SIZE(s.author) AS authors , concat_ws('\u003B',s.source.value) AS source,
|
||||
INSERT INTO ${stats_db_name}.software_tmp
|
||||
SELECT substr(s.id, 4) as id,
|
||||
s.title[0].value AS title,
|
||||
s.publisher.value AS publisher,
|
||||
CAST(NULL AS string) AS journal,
|
||||
s.dateofacceptance.value AS DATE,
|
||||
date_format(s.dateofacceptance.value, 'yyyy') AS YEAR,
|
||||
s.bestaccessright.classname AS bestlicence,
|
||||
s.embargoenddate.value AS embargo_end_date,
|
||||
FALSE AS delayed,
|
||||
SIZE(s.author) AS authors,
|
||||
concat_ws('\u003B', s.source.value) AS source,
|
||||
CASE WHEN SIZE(s.description) > 0 THEN TRUE ELSE FALSE END AS abstract,
|
||||
'software' as type
|
||||
'software' as type
|
||||
from ${openaire_db_name}.software s
|
||||
where s.datainfo.deletedbyinference=false;
|
||||
where s.datainfo.deletedbyinference = false;
|
||||
|
||||
-- Software_citations
|
||||
CREATE TABLE ${stats_db_name}.software_citations AS SELECT substr(s.id, 4) as id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS RESULT FROM ${openaire_db_name}.software s LATERAL VIEW explode(s.extrainfo) citations as citation where xpath_string(citation.value, "//citation/id[@type='openaire']/@value") !="" and s.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.software_citations AS
|
||||
SELECT substr(s.id, 4) as id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS RESULT
|
||||
FROM ${openaire_db_name}.software s
|
||||
LATERAL VIEW explode(s.extrainfo) citations as citation
|
||||
where xpath_string(citation.value, "//citation/id[@type='openaire']/@value") != ""
|
||||
and s.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.software_classifications AS SELECT substr(p.id, 4) AS id, instancetype.classname AS type FROM ${openaire_db_name}.software p LATERAL VIEW explode(p.instance.instancetype) instances AS instancetype where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.software_classifications AS
|
||||
SELECT substr(p.id, 4) AS id, instancetype.classname AS type
|
||||
FROM ${openaire_db_name}.software p
|
||||
LATERAL VIEW explode(p.instance.instancetype) instances AS instancetype
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.software_concepts AS SELECT substr(p.id, 4) AS id, contexts.context.id AS concept FROM ${openaire_db_name}.software p LATERAL VIEW explode(p.context) contexts AS context where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.software_concepts AS
|
||||
SELECT substr(p.id, 4) AS id, contexts.context.id AS concept
|
||||
FROM ${openaire_db_name}.software p
|
||||
LATERAL VIEW explode(p.context) contexts AS context
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.software_datasources AS SELECT p.id, CASE WHEN d.id IS NULL THEN 'other' ELSE p.datasource end as datasource FROM (SELECT substr(p.id, 4) AS id, substr(instances.instance.hostedby.key, 4) AS datasource
|
||||
FROM ${openaire_db_name}.software p LATERAL VIEW explode(p.instance) instances AS instance where p.datainfo.deletedbyinference=false) p LEFT OUTER JOIN
|
||||
(SELECT substr(d.id, 4) id FROM ${openaire_db_name}.datasource d WHERE d.datainfo.deletedbyinference=false) d ON p.datasource = d.id;
|
||||
CREATE TABLE ${stats_db_name}.software_datasources AS
|
||||
SELECT p.id, CASE WHEN d.id IS NULL THEN 'other' ELSE p.datasource end as datasource
|
||||
FROM (
|
||||
SELECT substr(p.id, 4) AS id, substr(instances.instance.hostedby.key, 4) AS datasource
|
||||
FROM ${openaire_db_name}.software p
|
||||
LATERAL VIEW explode(p.instance) instances AS instance
|
||||
where p.datainfo.deletedbyinference = false) p
|
||||
LEFT OUTER JOIN (
|
||||
SELECT substr(d.id, 4) id
|
||||
FROM ${openaire_db_name}.datasource d
|
||||
WHERE d.datainfo.deletedbyinference = false) d ON p.datasource = d.id;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.software_languages AS select substr(p.id, 4) AS id, p.language.classname AS language FROM ${openaire_db_name}.software p where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.software_languages AS
|
||||
select substr(p.id, 4) AS id, p.language.classname AS language
|
||||
FROM ${openaire_db_name}.software p
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.software_oids AS SELECT substr(p.id, 4) AS id, oids.ids AS oid FROM ${openaire_db_name}.software p LATERAL VIEW explode(p.originalid) oids AS ids where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.software_oids AS
|
||||
SELECT substr(p.id, 4) AS id, oids.ids AS oid
|
||||
FROM ${openaire_db_name}.software p
|
||||
LATERAL VIEW explode(p.originalid) oids AS ids
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.software_pids AS SELECT substr(p.id, 4) AS id, ppid.qualifier.classname AS type, ppid.value AS pid FROM ${openaire_db_name}.software p LATERAL VIEW explode(p.pid) pids AS ppid where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.software_pids AS
|
||||
SELECT substr(p.id, 4) AS id, ppid.qualifier.classname AS type, ppid.value AS pid
|
||||
FROM ${openaire_db_name}.software p
|
||||
LATERAL VIEW explode(p.pid) pids AS ppid
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.software_topics AS SELECT substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS type, subjects.subject.value AS topic FROM ${openaire_db_name}.software p LATERAL VIEW explode(p.subject) subjects AS subject where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.software_topics AS
|
||||
SELECT substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS type, subjects.subject.value AS topic
|
||||
FROM ${openaire_db_name}.software p
|
||||
LATERAL VIEW explode(p.subject) subjects AS subject
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.software_tmp COMPUTE STATISTICS;
|
||||
ANALYZE TABLE ${stats_db_name}.software_tmp COMPUTE STATISTICS FOR COLUMNS;
|
||||
|
|
|
|||
|
|
@ -5,32 +5,85 @@
|
|||
--------------------------------------------------------------------------------
|
||||
|
||||
-- Otherresearchproduct temporary table supporting updates
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_tmp ( id STRING, title STRING, publisher STRING, journal STRING, date STRING, year STRING, bestlicence STRING, embargo_end_date STRING, delayed BOOLEAN, authors INT, source STRING, abstract BOOLEAN, type STRING ) CLUSTERED BY (id) INTO 100 buckets stored AS orc tblproperties('transactional'='true');
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_tmp
|
||||
(
|
||||
id STRING,
|
||||
title STRING,
|
||||
publisher STRING,
|
||||
journal STRING,
|
||||
date STRING,
|
||||
year STRING,
|
||||
bestlicence STRING,
|
||||
embargo_end_date STRING,
|
||||
delayed BOOLEAN,
|
||||
authors INT,
|
||||
source STRING,
|
||||
abstract BOOLEAN,
|
||||
type STRING
|
||||
) CLUSTERED BY (id) INTO 100 buckets stored AS orc tblproperties ('transactional' = 'true');
|
||||
|
||||
INSERT INTO ${stats_db_name}.otherresearchproduct_tmp SELECT substr(o.id, 4) AS id, o.title[0].value AS title, o.publisher.value AS publisher, CAST(NULL AS string) AS journal,
|
||||
o.dateofacceptance.value AS DATE, date_format(o.dateofacceptance.value,'yyyy') AS year, o.bestaccessright.classname AS bestlicence,
|
||||
o.embargoenddate.value as embargo_end_date, FALSE AS delayed, SIZE(o.author) AS authors , concat_ws('\u003B',o.source.value) AS source,
|
||||
CASE WHEN SIZE(o.description) > 0 THEN TRUE ELSE FALSE END AS abstract,
|
||||
'other' AS type
|
||||
INSERT INTO ${stats_db_name}.otherresearchproduct_tmp
|
||||
SELECT substr(o.id, 4) AS id,
|
||||
o.title[0].value AS title,
|
||||
o.publisher.value AS publisher,
|
||||
CAST(NULL AS string) AS journal,
|
||||
o.dateofacceptance.value AS DATE,
|
||||
date_format(o.dateofacceptance.value, 'yyyy') AS year,
|
||||
o.bestaccessright.classname AS bestlicence,
|
||||
o.embargoenddate.value as embargo_end_date,
|
||||
FALSE AS delayed,
|
||||
SIZE(o.author) AS authors,
|
||||
concat_ws('\u003B', o.source.value) AS source,
|
||||
CASE WHEN SIZE(o.description) > 0 THEN TRUE ELSE FALSE END AS abstract,
|
||||
'other' AS type
|
||||
FROM ${openaire_db_name}.otherresearchproduct o
|
||||
WHERE o.datainfo.deletedbyinference=FALSE;
|
||||
WHERE o.datainfo.deletedbyinference = FALSE;
|
||||
|
||||
-- Otherresearchproduct_citations
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_citations AS SELECT substr(o.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS RESULT FROM ${openaire_db_name}.otherresearchproduct o LATERAL VIEW explode(o.extrainfo) citations AS citation WHERE xpath_string(citation.value, "//citation/id[@type='openaire']/@value") !="" and o.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_citations AS
|
||||
SELECT substr(o.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS RESULT
|
||||
FROM ${openaire_db_name}.otherresearchproduct o LATERAL VIEW explode(o.extrainfo) citations AS citation
|
||||
WHERE xpath_string(citation.value, "//citation/id[@type='openaire']/@value") != ""
|
||||
and o.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_classifications AS SELECT substr(p.id, 4) AS id, instancetype.classname AS type FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.instance.instancetype) instances AS instancetype where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_classifications AS
|
||||
SELECT substr(p.id, 4) AS id, instancetype.classname AS type
|
||||
FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.instance.instancetype) instances AS instancetype
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_concepts AS SELECT substr(p.id, 4) AS id, contexts.context.id AS concept FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.context) contexts AS context where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_concepts AS
|
||||
SELECT substr(p.id, 4) AS id, contexts.context.id AS concept
|
||||
FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.context) contexts AS context
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_datasources AS SELECT p.id, CASE WHEN d.id IS NULL THEN 'other' ELSE p.datasource END AS datasource FROM (SELECT substr(p.id, 4) AS id, substr(instances.instance.hostedby.key, 4) AS datasource from ${openaire_db_name}.otherresearchproduct p lateral view explode(p.instance) instances as instance where p.datainfo.deletedbyinference=false) p LEFT OUTER JOIN(SELECT substr(d.id, 4) id from ${openaire_db_name}.datasource d WHERE d.datainfo.deletedbyinference=false) d on p.datasource = d.id;
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_datasources AS
|
||||
SELECT p.id, CASE WHEN d.id IS NULL THEN 'other' ELSE p.datasource END AS datasource
|
||||
FROM (SELECT substr(p.id, 4) AS id, substr(instances.instance.hostedby.key, 4) AS datasource
|
||||
from ${openaire_db_name}.otherresearchproduct p lateral view explode(p.instance) instances as instance
|
||||
where p.datainfo.deletedbyinference = false) p
|
||||
LEFT OUTER JOIN(SELECT substr(d.id, 4) id
|
||||
from ${openaire_db_name}.datasource d
|
||||
WHERE d.datainfo.deletedbyinference = false) d on p.datasource = d.id;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_languages AS SELECT substr(p.id, 4) AS id, p.language.classname AS language FROM ${openaire_db_name}.otherresearchproduct p where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_languages AS
|
||||
SELECT substr(p.id, 4) AS id, p.language.classname AS language
|
||||
FROM ${openaire_db_name}.otherresearchproduct p
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_oids AS SELECT substr(p.id, 4) AS id, oids.ids AS oid FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.originalid) oids AS ids where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_oids AS
|
||||
SELECT substr(p.id, 4) AS id, oids.ids AS oid
|
||||
FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.originalid) oids AS ids
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_pids AS SELECT substr(p.id, 4) AS id, ppid.qualifier.classname AS type, ppid.value AS pid FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.pid) pids AS ppid where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_pids AS
|
||||
SELECT substr(p.id, 4) AS id, ppid.qualifier.classname AS type, ppid.value AS pid
|
||||
FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.pid) pids AS ppid
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_topics AS SELECT substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS type, subjects.subject.value AS topic FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.subject) subjects AS subject where p.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.otherresearchproduct_topics AS
|
||||
SELECT substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS type, subjects.subject.value AS topic
|
||||
FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.subject) subjects AS subject
|
||||
where p.datainfo.deletedbyinference = false;
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.otherresearchproduct_tmp COMPUTE STATISTICS;
|
||||
ANALYZE TABLE ${stats_db_name}.otherresearchproduct_tmp COMPUTE STATISTICS FOR COLUMNS;
|
||||
|
|
|
|||
|
|
@ -1,28 +1,75 @@
|
|||
-- noinspection SqlNoDataSourceInspectionForFile
|
||||
|
||||
------------------------------------------------------
|
||||
------------------------------------------------------
|
||||
-- Project table/view and Project related tables/views
|
||||
------------------------------------------------------
|
||||
------------------------------------------------------
|
||||
-- Project_oids Table
|
||||
CREATE TABLE ${stats_db_name}.project_oids AS SELECT substr(p.id, 4) AS id, oids.ids AS oid FROM ${openaire_db_name}.project p LATERAL VIEW explode(p.originalid) oids AS ids;
|
||||
CREATE TABLE ${stats_db_name}.project_oids AS
|
||||
SELECT substr(p.id, 4) AS id, oids.ids AS oid
|
||||
FROM ${openaire_db_name}.project p LATERAL VIEW explode(p.originalid) oids AS ids;
|
||||
CREATE TABLE ${stats_db_name}.project_organizations AS
|
||||
SELECT substr(r.source, 4) AS id, substr(r.target, 4) AS organization
|
||||
from ${openaire_db_name}.relation r
|
||||
WHERE r.reltype = 'projectOrganization'
|
||||
and r.datainfo.deletedbyinference = false;
|
||||
|
||||
-- Project_organizations Table
|
||||
CREATE TABLE ${stats_db_name}.project_organizations AS SELECT substr(r.source, 4) AS id, substr(r.target, 4) AS organization from ${openaire_db_name}.relation r WHERE r.reltype='projectOrganization' and r.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.project_results AS
|
||||
SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS result
|
||||
FROM ${openaire_db_name}.relation r
|
||||
WHERE r.reltype = 'resultProject'
|
||||
and r.datainfo.deletedbyinference = false;
|
||||
|
||||
-- Project_results Table
|
||||
CREATE TABLE ${stats_db_name}.project_results AS SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS result FROM ${openaire_db_name}.relation r WHERE r.reltype='resultProject' and r.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.project_tmp
|
||||
(
|
||||
id STRING,
|
||||
acronym STRING,
|
||||
title STRING,
|
||||
funder STRING,
|
||||
funding_lvl0 STRING,
|
||||
funding_lvl1 STRING,
|
||||
funding_lvl2 STRING,
|
||||
ec39 STRING,
|
||||
type STRING,
|
||||
startdate STRING,
|
||||
enddate STRING,
|
||||
start_year INT,
|
||||
end_year INT,
|
||||
duration INT,
|
||||
haspubs STRING,
|
||||
numpubs INT,
|
||||
daysforlastpub INT,
|
||||
delayedpubs INT,
|
||||
callidentifier STRING,
|
||||
code STRING
|
||||
) CLUSTERED BY (id) INTO 100 buckets stored AS orc tblproperties ('transactional' = 'true');
|
||||
|
||||
-- Project table
|
||||
----------------
|
||||
-- Creating and populating temporary Project table
|
||||
CREATE TABLE ${stats_db_name}.project_tmp (id STRING, acronym STRING, title STRING, funder STRING, funding_lvl0 STRING, funding_lvl1 STRING, funding_lvl2 STRING, ec39 STRING, type STRING, startdate STRING, enddate STRING, start_year INT, end_year INT, duration INT, haspubs STRING, numpubs INT, daysforlastpub INT, delayedpubs INT, callidentifier STRING, code STRING) CLUSTERED BY (id) INTO 100 buckets stored AS orc tblproperties('transactional'='true');
|
||||
|
||||
INSERT INTO ${stats_db_name}.project_tmp SELECT substr(p.id, 4) AS id, p.acronym.value AS acronym, p.title.value AS title, xpath_string(p.fundingtree[0].value, '//funder/name') AS funder, xpath_string(p.fundingtree[0].value, '//funding_level_0/name') AS funding_lvl0, xpath_string(p.fundingtree[0].value, '//funding_level_1/name') AS funding_lvl1, xpath_string(p.fundingtree[0].value, '//funding_level_2/name') AS funding_lvl2, p.ecsc39.value AS ec39, p.contracttype.classname AS type, p.startdate.value AS startdate, p.enddate.value AS enddate, year(p.startdate.value) AS start_year, year(p.enddate.value) AS end_year, CAST(MONTHS_BETWEEN(p.enddate.value, p.startdate.value) AS INT) AS duration, 'no' AS haspubs, 0 AS numpubs, 0 AS daysforlastpub, 0 AS delayedpubs, p.callidentifier.value AS callidentifier, p.code.value AS code FROM ${openaire_db_name}.project p WHERE p.datainfo.deletedbyinference=false;
|
||||
INSERT INTO ${stats_db_name}.project_tmp
|
||||
SELECT substr(p.id, 4) AS id,
|
||||
p.acronym.value AS acronym,
|
||||
p.title.value AS title,
|
||||
xpath_string(p.fundingtree[0].value, '//funder/name') AS funder,
|
||||
xpath_string(p.fundingtree[0].value, '//funding_level_0/name') AS funding_lvl0,
|
||||
xpath_string(p.fundingtree[0].value, '//funding_level_1/name') AS funding_lvl1,
|
||||
xpath_string(p.fundingtree[0].value, '//funding_level_2/name') AS funding_lvl2,
|
||||
p.ecsc39.value AS ec39,
|
||||
p.contracttype.classname AS type,
|
||||
p.startdate.value AS startdate,
|
||||
p.enddate.value AS enddate,
|
||||
year(p.startdate.value) AS start_year,
|
||||
year(p.enddate.value) AS end_year,
|
||||
CAST(MONTHS_BETWEEN(p.enddate.value, p.startdate.value) AS INT) AS duration,
|
||||
'no' AS haspubs,
|
||||
0 AS numpubs,
|
||||
0 AS daysforlastpub,
|
||||
0 AS delayedpubs,
|
||||
p.callidentifier.value AS callidentifier,
|
||||
p.code.value AS code
|
||||
FROM ${openaire_db_name}.project p
|
||||
WHERE p.datainfo.deletedbyinference = false;
|
||||
|
||||
create table ${stats_db_name}.funder as
|
||||
select distinct xpath_string(fund, '//funder/id') as id, xpath_string(fund, '//funder/name') as name, xpath_string(fund, '//funder/shortname') as shortname
|
||||
select distinct xpath_string(fund, '//funder/id') as id,
|
||||
xpath_string(fund, '//funder/name') as name,
|
||||
xpath_string(fund, '//funder/shortname') as shortname
|
||||
from ${openaire_db_name}.project p lateral view explode(p.fundingtree.value) fundingtree as fund
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.project_oids COMPUTE STATISTICS;
|
||||
|
|
|
|||
|
|
@ -5,28 +5,135 @@
|
|||
----------------------------------------------------
|
||||
|
||||
-- Views on temporary tables that should be re-created in the end
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result as SELECT *, bestlicence AS access_mode FROM ${stats_db_name}.publication_tmp UNION ALL SELECT *,bestlicence AS access_mode FROM ${stats_db_name}.software_tmp UNION ALL SELECT *,bestlicence AS access_mode FROM ${stats_db_name}.dataset_tmp UNION ALL SELECT *,bestlicence AS access_mode FROM ${stats_db_name}.otherresearchproduct_tmp;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result as
|
||||
SELECT *, bestlicence AS access_mode
|
||||
FROM ${stats_db_name}.publication_tmp
|
||||
UNION ALL
|
||||
SELECT *, bestlicence AS access_mode
|
||||
FROM ${stats_db_name}.software_tmp
|
||||
UNION ALL
|
||||
SELECT *, bestlicence AS access_mode
|
||||
FROM ${stats_db_name}.dataset_tmp
|
||||
UNION ALL
|
||||
SELECT *, bestlicence AS access_mode
|
||||
FROM ${stats_db_name}.otherresearchproduct_tmp;
|
||||
|
||||
-- Views on final tables
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_datasources AS SELECT * FROM ${stats_db_name}.publication_datasources UNION ALL SELECT * FROM ${stats_db_name}.software_datasources UNION ALL SELECT * FROM ${stats_db_name}.dataset_datasources UNION ALL SELECT * FROM ${stats_db_name}.otherresearchproduct_datasources;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_datasources AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.publication_datasources
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.software_datasources
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.dataset_datasources
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.otherresearchproduct_datasources;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_citations AS SELECT * FROM ${stats_db_name}.publication_citations UNION ALL SELECT * FROM ${stats_db_name}.software_citations UNION ALL SELECT * FROM ${stats_db_name}.dataset_citations UNION ALL SELECT * FROM ${stats_db_name}.otherresearchproduct_citations;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_citations AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.publication_citations
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.software_citations
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.dataset_citations
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.otherresearchproduct_citations;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_classifications AS SELECT * FROM ${stats_db_name}.publication_classifications UNION ALL SELECT * FROM ${stats_db_name}.software_classifications UNION ALL SELECT * FROM ${stats_db_name}.dataset_classifications UNION ALL SELECT * FROM ${stats_db_name}.otherresearchproduct_classifications;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_classifications AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.publication_classifications
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.software_classifications
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.dataset_classifications
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.otherresearchproduct_classifications;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_concepts AS SELECT * FROM ${stats_db_name}.publication_concepts UNION ALL SELECT * FROM ${stats_db_name}.software_concepts UNION ALL SELECT * FROM ${stats_db_name}.dataset_concepts UNION ALL SELECT * FROM ${stats_db_name}.otherresearchproduct_concepts;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_concepts AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.publication_concepts
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.software_concepts
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.dataset_concepts
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.otherresearchproduct_concepts;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_languages AS SELECT * FROM ${stats_db_name}.publication_languages UNION ALL SELECT * FROM ${stats_db_name}.software_languages UNION ALL SELECT * FROM ${stats_db_name}.dataset_languages UNION ALL SELECT * FROM ${stats_db_name}.otherresearchproduct_languages;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_languages AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.publication_languages
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.software_languages
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.dataset_languages
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.otherresearchproduct_languages;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_oids AS SELECT * FROM ${stats_db_name}.publication_oids UNION ALL SELECT * FROM ${stats_db_name}.software_oids UNION ALL SELECT * FROM ${stats_db_name}.dataset_oids UNION ALL SELECT * FROM ${stats_db_name}.otherresearchproduct_oids;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_oids AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.publication_oids
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.software_oids
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.dataset_oids
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.otherresearchproduct_oids;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_pids AS SELECT * FROM ${stats_db_name}.publication_pids UNION ALL SELECT * FROM ${stats_db_name}.software_pids UNION ALL SELECT * FROM ${stats_db_name}.dataset_pids UNION ALL SELECT * FROM ${stats_db_name}.otherresearchproduct_pids;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_pids AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.publication_pids
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.software_pids
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.dataset_pids
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.otherresearchproduct_pids;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_topics AS SELECT * FROM ${stats_db_name}.publication_topics UNION ALL SELECT * FROM ${stats_db_name}.software_topics UNION ALL SELECT * FROM ${stats_db_name}.dataset_topics UNION ALL SELECT * FROM ${stats_db_name}.otherresearchproduct_topics;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.result_topics AS
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.publication_topics
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.software_topics
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.dataset_topics
|
||||
UNION ALL
|
||||
SELECT *
|
||||
FROM ${stats_db_name}.otherresearchproduct_topics;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.result_organization AS SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS organization FROM ${openaire_db_name}.relation r WHERE r.reltype='resultOrganization' and r.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.result_organization AS
|
||||
SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS organization
|
||||
FROM ${openaire_db_name}.relation r
|
||||
WHERE r.reltype = 'resultOrganization'
|
||||
and r.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE TABLE ${stats_db_name}.result_projects AS select pr.result AS id, pr.id AS project, datediff(p.enddate, p.startdate) AS daysfromend FROM ${stats_db_name}.result r JOIN ${stats_db_name}.project_results pr ON r.id=pr.result JOIN ${stats_db_name}.project_tmp p ON p.id=pr.id;
|
||||
CREATE TABLE ${stats_db_name}.result_projects AS
|
||||
select pr.result AS id, pr.id AS project, datediff(p.enddate, p.startdate) AS daysfromend
|
||||
FROM ${stats_db_name}.result r
|
||||
JOIN ${stats_db_name}.project_results pr ON r.id = pr.result
|
||||
JOIN ${stats_db_name}.project_tmp p ON p.id = pr.id;
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.result_organization COMPUTE STATISTICS;
|
||||
ANALYZE TABLE ${stats_db_name}.result_organization COMPUTE STATISTICS FOR COLUMNS;
|
||||
|
|
|
|||
|
|
@ -5,55 +5,99 @@
|
|||
-- Datasource table/view and Datasource related tables/views
|
||||
------------------------------------------------------------
|
||||
------------------------------------------------------------
|
||||
|
||||
-- Datasource table creation & update
|
||||
-------------------------------------
|
||||
-- Creating and populating temporary datasource table
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.datasource_tmp;
|
||||
CREATE TABLE ${stats_db_name}.datasource_tmp(`id` string, `name` STRING, `type` STRING, `dateofvalidation` STRING, `yearofvalidation` string, `harvested` BOOLEAN, `piwik_id` INT, `latitude` STRING, `longitude`STRING, `websiteurl` STRING, `compatibility` STRING) CLUSTERED BY (id) INTO 100 buckets stored AS orc tblproperties('transactional'='true');
|
||||
CREATE TABLE ${stats_db_name}.datasource_tmp
|
||||
(
|
||||
`id` string,
|
||||
`name` STRING,
|
||||
`type` STRING,
|
||||
`dateofvalidation` STRING,
|
||||
`yearofvalidation` string,
|
||||
`harvested` BOOLEAN,
|
||||
`piwik_id` INT,
|
||||
`latitude` STRING,
|
||||
`longitude` STRING,
|
||||
`websiteurl` STRING,
|
||||
`compatibility` STRING
|
||||
) CLUSTERED BY (id) INTO 100 buckets stored AS orc tblproperties ('transactional' = 'true');
|
||||
|
||||
-- Insert statement that takes into account the piwik_id of the openAIRE graph
|
||||
INSERT INTO ${stats_db_name}.datasource_tmp
|
||||
SELECT substr(d1.id, 4) AS id, officialname.value AS name,
|
||||
datasourcetype.classname AS type, dateofvalidation.value AS dateofvalidation, date_format(d1.dateofvalidation.value,'yyyy') AS yearofvalidation,
|
||||
FALSE AS harvested,
|
||||
CASE WHEN d2.piwik_id IS NULL THEN 0 ELSE d2.piwik_id END AS piwik_id,
|
||||
d1.latitude.value AS latitude, d1.longitude.value AS longitude,
|
||||
d1.websiteurl.value AS websiteurl, d1.openairecompatibility.classid AS compatibility
|
||||
SELECT substr(d1.id, 4) AS id,
|
||||
officialname.value AS name,
|
||||
datasourcetype.classname AS type,
|
||||
dateofvalidation.value AS dateofvalidation,
|
||||
date_format(d1.dateofvalidation.value, 'yyyy') AS yearofvalidation,
|
||||
FALSE AS harvested,
|
||||
CASE WHEN d2.piwik_id IS NULL THEN 0 ELSE d2.piwik_id END AS piwik_id,
|
||||
d1.latitude.value AS latitude,
|
||||
d1.longitude.value AS longitude,
|
||||
d1.websiteurl.value AS websiteurl,
|
||||
d1.openairecompatibility.classid AS compatibility
|
||||
FROM ${openaire_db_name}.datasource d1
|
||||
LEFT OUTER JOIN
|
||||
(SELECT id, split(originalidd, '\\:')[1] as piwik_id
|
||||
FROM ${openaire_db_name}.datasource
|
||||
LATERAL VIEW EXPLODE(originalid) temp AS originalidd
|
||||
WHERE originalidd like "piwik:%") AS d2
|
||||
ON d1.id = d2.id
|
||||
WHERE d1.datainfo.deletedbyinference=FALSE;
|
||||
LEFT OUTER JOIN
|
||||
(SELECT id, split(originalidd, '\\:')[1] as piwik_id
|
||||
FROM ${openaire_db_name}.datasource
|
||||
LATERAL VIEW EXPLODE(originalid) temp AS originalidd
|
||||
WHERE originalidd like "piwik:%") AS d2
|
||||
ON d1.id = d2.id
|
||||
WHERE d1.datainfo.deletedbyinference = FALSE;
|
||||
|
||||
-- Updating temporary table with everything that is not based on results -> This is done with the following "dual" table.
|
||||
-- Creating a temporary dual table that will be removed after the following insert
|
||||
CREATE TABLE ${stats_db_name}.dual(dummy CHAR(1));
|
||||
INSERT INTO ${stats_db_name}.dual VALUES('X');
|
||||
INSERT INTO ${stats_db_name}.datasource_tmp (`id`, `name`, `type`, `dateofvalidation`, `yearofvalidation`, `harvested`, `piwik_id`, `latitude`, `longitude`, `websiteurl`, `compatibility`)
|
||||
SELECT 'other', 'Other', 'Repository', NULL, NULL, false, 0, NULL, NULL, NULL, 'unknown' FROM ${stats_db_name}.dual WHERE 'other' not in (SELECT id FROM ${stats_db_name}.datasource_tmp WHERE name='Unknown Repository');
|
||||
CREATE TABLE ${stats_db_name}.dual
|
||||
(
|
||||
dummy CHAR(1)
|
||||
);
|
||||
INSERT INTO ${stats_db_name}.dual
|
||||
VALUES ('X');
|
||||
INSERT INTO ${stats_db_name}.datasource_tmp (`id`, `name`, `type`, `dateofvalidation`, `yearofvalidation`, `harvested`,
|
||||
`piwik_id`, `latitude`, `longitude`, `websiteurl`, `compatibility`)
|
||||
SELECT 'other',
|
||||
'Other',
|
||||
'Repository',
|
||||
NULL,
|
||||
NULL,
|
||||
false,
|
||||
0,
|
||||
NULL,
|
||||
NULL,
|
||||
NULL,
|
||||
'unknown'
|
||||
FROM ${stats_db_name}.dual
|
||||
WHERE 'other' not in (SELECT id FROM ${stats_db_name}.datasource_tmp WHERE name = 'Unknown Repository');
|
||||
DROP TABLE ${stats_db_name}.dual;
|
||||
|
||||
UPDATE ${stats_db_name}.datasource_tmp SET name='Other' WHERE name='Unknown Repository';
|
||||
UPDATE ${stats_db_name}.datasource_tmp SET yearofvalidation=null WHERE yearofvalidation='-1';
|
||||
UPDATE ${stats_db_name}.datasource_tmp
|
||||
SET name='Other'
|
||||
WHERE name = 'Unknown Repository';
|
||||
UPDATE ${stats_db_name}.datasource_tmp
|
||||
SET yearofvalidation=null
|
||||
WHERE yearofvalidation = '-1';
|
||||
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.datasource_languages;
|
||||
CREATE TABLE ${stats_db_name}.datasource_languages AS SELECT substr(d.id, 4) AS id, langs.languages AS language FROM ${openaire_db_name}.datasource d LATERAL VIEW explode(d.odlanguages.value) langs AS languages;
|
||||
CREATE TABLE ${stats_db_name}.datasource_languages AS
|
||||
SELECT substr(d.id, 4) AS id, langs.languages AS language
|
||||
FROM ${openaire_db_name}.datasource d LATERAL VIEW explode(d.odlanguages.value) langs AS languages;
|
||||
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.datasource_oids;
|
||||
CREATE TABLE ${stats_db_name}.datasource_oids AS SELECT substr(d.id, 4) AS id, oids.ids AS oid FROM ${openaire_db_name}.datasource d LATERAL VIEW explode(d.originalid) oids AS ids;
|
||||
CREATE TABLE ${stats_db_name}.datasource_oids AS
|
||||
SELECT substr(d.id, 4) AS id, oids.ids AS oid
|
||||
FROM ${openaire_db_name}.datasource d LATERAL VIEW explode(d.originalid) oids AS ids;
|
||||
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.datasource_organizations;
|
||||
CREATE TABLE ${stats_db_name}.datasource_organizations AS SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS organization FROM ${openaire_db_name}.relation r WHERE r.reltype='datasourceOrganization' and r.datainfo.deletedbyinference=false;
|
||||
CREATE TABLE ${stats_db_name}.datasource_organizations AS
|
||||
SELECT substr(r.target, 4) AS id, substr(r.source, 4) AS organization
|
||||
FROM ${openaire_db_name}.relation r
|
||||
WHERE r.reltype = 'datasourceOrganization'
|
||||
and r.datainfo.deletedbyinference = false;
|
||||
|
||||
-- datasource sources:
|
||||
-- where the datasource info have been collected from.
|
||||
create table if not exists ${stats_db_name}.datasource_sources AS select substr(d.id,4) as id, substr(cf.key, 4) as datasource from ${openaire_db_name}.datasource d lateral view explode(d.collectedfrom) cfrom as cf where d.datainfo.deletedbyinference=false;
|
||||
create table if not exists ${stats_db_name}.datasource_sources AS
|
||||
select substr(d.id, 4) as id, substr(cf.key, 4) as datasource
|
||||
from ${openaire_db_name}.datasource d lateral view explode(d.collectedfrom) cfrom as cf
|
||||
where d.datainfo.deletedbyinference = false;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.datasource_results AS SELECT datasource AS id, id AS result FROM ${stats_db_name}.result_datasources;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.datasource_results AS
|
||||
SELECT datasource AS id, id AS result
|
||||
FROM ${stats_db_name}.result_datasources;
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.datasource_tmp COMPUTE STATISTICS;
|
||||
ANALYZE TABLE ${stats_db_name}.datasource_tmp COMPUTE STATISTICS FOR COLUMNS;
|
||||
|
|
|
|||
|
|
@ -3,13 +3,21 @@
|
|||
-- Organization table/view and Organization related tables/views
|
||||
----------------------------------------------------------------
|
||||
----------------------------------------------------------------
|
||||
DROP TABLE IF EXISTS ${stats_db_name}.organization;
|
||||
CREATE TABLE IF NOT EXISTS ${stats_db_name}.organization AS SELECT substr(o.id, 4) as id, o.legalname.value as name, o.legalshortname.value as legalshortname, o.country.classid as country
|
||||
FROM ${openaire_db_name}.organization o WHERE o.datainfo.deletedbyinference=FALSE;
|
||||
CREATE TABLE IF NOT EXISTS ${stats_db_name}.organization AS
|
||||
SELECT substr(o.id, 4) as id,
|
||||
o.legalname.value as name,
|
||||
o.legalshortname.value as legalshortname,
|
||||
o.country.classid as country
|
||||
FROM ${openaire_db_name}.organization o
|
||||
WHERE o.datainfo.deletedbyinference = FALSE;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.organization_datasources AS SELECT organization AS id, id AS datasource FROM ${stats_db_name}.datasource_organizations;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.organization_datasources AS
|
||||
SELECT organization AS id, id AS datasource
|
||||
FROM ${stats_db_name}.datasource_organizations;
|
||||
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.organization_projects AS SELECT id AS project, organization as id FROM ${stats_db_name}.project_organizations;
|
||||
CREATE OR REPLACE VIEW ${stats_db_name}.organization_projects AS
|
||||
SELECT id AS project, organization as id
|
||||
FROM ${stats_db_name}.project_organizations;
|
||||
|
||||
ANALYZE TABLE ${stats_db_name}.organization COMPUTE STATISTICS;
|
||||
ANALYZE TABLE ${stats_db_name}.organization COMPUTE STATISTICS FOR COLUMNS;
|
||||
Loading…
Reference in New Issue