From 0948536614d1c1632a93ca514fe5b2e0215ec622 Mon Sep 17 00:00:00 2001 From: antleb Date: Wed, 2 Dec 2020 15:41:56 +0200 Subject: [PATCH] initial implementation of the promote wf --- dhp-workflows/dhp-stats-promote/pom.xml | 32 ++ .../graph/stats/oozie_app/config-default.xml | 34 ++ .../oa/graph/stats/oozie_app/impala-shell.sh | 18 ++ .../oa/graph/stats/oozie_app/promoteCache.sh | 4 + .../graph/stats/oozie_app/scripts/step1.sql | 8 + .../graph/stats/oozie_app/scripts/step10.sql | 21 ++ .../graph/stats/oozie_app/scripts/step11.sql | 44 +++ .../graph/stats/oozie_app/scripts/step12.sql | 38 +++ .../graph/stats/oozie_app/scripts/step13.sql | 59 ++++ .../graph/stats/oozie_app/scripts/step14.sql | 49 +++ .../graph/stats/oozie_app/scripts/step15.sql | 36 +++ .../graph/stats/oozie_app/scripts/step16.sql | 80 +++++ .../stats/oozie_app/scripts/step16_5.sql | 55 ++++ .../stats/oozie_app/scripts/step16_6.sql | 32 ++ .../graph/stats/oozie_app/scripts/step17.sql | 207 ++++++++++++ .../graph/stats/oozie_app/scripts/step18.sql | 8 + .../graph/stats/oozie_app/scripts/step19.sql | 8 + .../graph/stats/oozie_app/scripts/step2.sql | 44 +++ .../graph/stats/oozie_app/scripts/step3.sql | 36 +++ .../graph/stats/oozie_app/scripts/step4.sql | 36 +++ .../graph/stats/oozie_app/scripts/step5.sql | 36 +++ .../graph/stats/oozie_app/scripts/step6.sql | 30 ++ .../graph/stats/oozie_app/scripts/step7.sql | 31 ++ .../graph/stats/oozie_app/scripts/step8.sql | 58 ++++ .../graph/stats/oozie_app/scripts/step9.sql | 12 + .../dhp/oa/graph/stats/oozie_app/workflow.xml | 302 ++++++++++++++++++ 26 files changed, 1318 insertions(+) create mode 100644 dhp-workflows/dhp-stats-promote/pom.xml create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/config-default.xml create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/impala-shell.sh create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/promoteCache.sh create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step1.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step10.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step11.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step12.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step13.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step14.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_5.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_6.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step17.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step18.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step19.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step2.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step3.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step4.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step5.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step6.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step8.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step9.sql create mode 100644 dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml diff --git a/dhp-workflows/dhp-stats-promote/pom.xml b/dhp-workflows/dhp-stats-promote/pom.xml new file mode 100644 index 000000000..c64c2f58e --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/pom.xml @@ -0,0 +1,32 @@ + + + + dhp-workflows + eu.dnetlib.dhp + 1.2.4-SNAPSHOT + + 4.0.0 + dhp-stats-promote + + + org.apache.spark + spark-core_2.11 + + + org.apache.spark + spark-sql_2.11 + + + + + + pl.project13.maven + git-commit-id-plugin + 2.1.11 + + false + + + + + diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/config-default.xml b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/config-default.xml new file mode 100644 index 000000000..9331d4ac5 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/config-default.xml @@ -0,0 +1,34 @@ + + + jobTracker + ${jobTracker} + + + nameNode + ${nameNode} + + + oozie.use.system.libpath + true + + + oozie.action.sharelib.for.spark + spark2 + + + hive_metastore_uris + thrift://iis-cdh5-test-m3.ocean.icm.edu.pl:9083 + + + hive_jdbc_url + jdbc:hive2://iis-cdh5-test-m3.ocean.icm.edu.pl:10000 + + + oozie.wf.workflow.notification.url + {serviceUrl}/v1/oozieNotification/jobUpdate?jobId=$jobId%26status=$status + + + stats_tool_api_url + ${stats_tool_api_url} + + \ No newline at end of file diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/impala-shell.sh b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/impala-shell.sh new file mode 100644 index 000000000..70112dc7b --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/impala-shell.sh @@ -0,0 +1,18 @@ +export PYTHON_EGG_CACHE=/home/$(whoami)/.python-eggs +export link_folder=/tmp/impala-shell-python-egg-cache-$(whoami) +if ! [ -L $link_folder ] +then + rm -Rf "$link_folder" + ln -sfn ${PYTHON_EGG_CACHE}${link_folder} ${link_folder} +fi + +echo "Getting file from " $3 +hdfs dfs -copyToLocal $3 + +echo "Running impala shell make the new database visible" +impala-shell -q "INVALIDATE METADATA;" + +echo "Running impala shell to compute new table stats" +impala-shell -d $1 -f $2 +echo "Impala shell finished" +rm $2 diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/promoteCache.sh b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/promoteCache.sh new file mode 100644 index 000000000..2d28377fb --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/promoteCache.sh @@ -0,0 +1,4 @@ +#!/usr/bin/env bash + +curl --request GET $1/cache/promoteCache + diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step1.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step1.sql new file mode 100644 index 000000000..9697a1dc8 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step1.sql @@ -0,0 +1,8 @@ +-------------------------------------------------------------- +-------------------------------------------------------------- +-- Stats database creation +-------------------------------------------------------------- +-------------------------------------------------------------- + +DROP database IF EXISTS ${stats_db_name} CASCADE; +CREATE database ${stats_db_name}; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step10.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step10.sql new file mode 100644 index 000000000..46ff295f4 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step10.sql @@ -0,0 +1,21 @@ +------------------------------------------------------------------------------------------------ +------------------------------------------------------------------------------------------------ +-- 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; + + +------------------------------------------------------------------------------------------------ +------------------------------------------------------------------------------------------------ +-- Creation date of the database +------------------------------------------------------------------------------------------------ +------------------------------------------------------------------------------------------------ +create table ${stats_db_name}.creation_date as select date_format(current_date(), 'dd-MM-yyyy') as date; \ No newline at end of file diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step11.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step11.sql new file mode 100644 index 000000000..13e141459 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step11.sql @@ -0,0 +1,44 @@ +---------------------------------------------------------------- +---------------------------------------------------------------- +-- Post processing - Updates on main tables +---------------------------------------------------------------- +---------------------------------------------------------------- + +--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); + +-- 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'); + +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 +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' + 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 + 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); + +-- 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); + +-- 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); + +-- 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); + +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; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step12.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step12.sql new file mode 100644 index 000000000..25439852e --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step12.sql @@ -0,0 +1,38 @@ +------------------------------------------------------------------------------------------------------ +-- Creating parquet tables from the updated temporary tables and removing unnecessary temporary tables +------------------------------------------------------------------------------------------------------ + +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; + +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; + +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; + +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; + +DROP TABLE ${stats_db_name}.project_tmp; +DROP TABLE ${stats_db_name}.datasource_tmp; +DROP TABLE ${stats_db_name}.publication_tmp; +DROP TABLE ${stats_db_name}.dataset_tmp; +DROP TABLE ${stats_db_name}.software_tmp; +DROP TABLE ${stats_db_name}.otherresearchproduct_tmp; + +---------------------------------------------- +-- Re-creating views from final parquet tables +--------------------------------------------- + +-- 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; + + +------------------------------------------------------------------------------- +-- 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; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step13.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step13.sql new file mode 100644 index 000000000..795770313 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step13.sql @@ -0,0 +1,59 @@ +------------------------------------------------------ +------------------------------------------------------ +-- Additional relations +-- +-- Sources related tables/views +------------------------------------------------------ +------------------------------------------------------ +CREATE TABLE IF NOT EXISTS ${stats_db_name}.publication_sources 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(datasource, 4) as datasource +from ${openaire_db_name}.publication p lateral view explode(p.collectedfrom.key) c as datasource) 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 IF NOT EXISTS ${stats_db_name}.dataset_sources 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(datasource, 4) as datasource +from ${openaire_db_name}.dataset p lateral view explode(p.collectedfrom.key) c as datasource) 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 IF NOT EXISTS ${stats_db_name}.software_sources 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(datasource, 4) as datasource +from ${openaire_db_name}.software p lateral view explode(p.collectedfrom.key) c as datasource) 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 IF NOT EXISTS ${stats_db_name}.otherresearchproduct_sources 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(datasource, 4) as datasource +from ${openaire_db_name}.otherresearchproduct p lateral view explode(p.collectedfrom.key) c as datasource) 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 VIEW IF NOT EXISTS ${stats_db_name}.result_sources AS +SELECT * FROM ${stats_db_name}.publication_sources +UNION ALL +SELECT * FROM ${stats_db_name}.dataset_sources +UNION ALL +SELECT * FROM ${stats_db_name}.software_sources +UNION ALL +SELECT * FROM ${stats_db_name}.otherresearchproduct_sources; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step14.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step14.sql new file mode 100644 index 000000000..4a56b5d68 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step14.sql @@ -0,0 +1,49 @@ +------------------------------------------------------ +------------------------------------------------------ +-- Additional relations +-- +-- Licences related tables/views +------------------------------------------------------ +------------------------------------------------------ +CREATE TABLE IF NOT EXISTS ${stats_db_name}.publication_licenses AS +SELECT substr(p.id, 4) as id, licenses.value as type +from ${openaire_db_name}.publication p LATERAL VIEW explode(p.instance.license) instances as licenses +where licenses.value is not null and licenses.value != '' and p.datainfo.deletedbyinference=false; + +CREATE TABLE IF NOT EXISTS ${stats_db_name}.dataset_licenses AS +SELECT substr(p.id, 4) as id, licenses.value as type +from ${openaire_db_name}.dataset p LATERAL VIEW explode(p.instance.license) instances as licenses +where licenses.value is not null and licenses.value != '' and p.datainfo.deletedbyinference=false; + +CREATE TABLE IF NOT EXISTS ${stats_db_name}.software_licenses AS +SELECT substr(p.id, 4) as id, licenses.value as type +from ${openaire_db_name}.software p LATERAL VIEW explode(p.instance.license) instances as licenses +where licenses.value is not null and licenses.value != '' and p.datainfo.deletedbyinference=false; + +CREATE TABLE IF NOT EXISTS ${stats_db_name}.otherresearchproduct_licenses AS +SELECT substr(p.id, 4) as id, licenses.value as type +from ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.instance.license) instances as licenses +where licenses.value is not null and licenses.value != '' and p.datainfo.deletedbyinference=false; + +CREATE VIEW IF NOT EXISTS ${stats_db_name}.result_licenses AS +SELECT * FROM ${stats_db_name}.publication_licenses +UNION ALL +SELECT * FROM ${stats_db_name}.dataset_licenses +UNION ALL +SELECT * FROM ${stats_db_name}.software_licenses +UNION ALL +SELECT * FROM ${stats_db_name}.otherresearchproduct_licenses; + +CREATE TABLE IF NOT EXISTS ${stats_db_name}.organization_pids AS +select substr(o.id, 4) as id, ppid.qualifier.classname as type, ppid.value as pid +from ${openaire_db_name}.organization o lateral view explode(o.pid) pids as ppid; + +CREATE TABLE IF NOT EXISTS ${stats_db_name}.organization_sources as +SELECT o.id, case when d.id is null then 'other' else o.datasource end as datasource +FROM ( + SELECT substr(o.id, 4) as id, substr(instances.instance.key, 4) as datasource + from ${openaire_db_name}.organization o lateral view explode(o.collectedfrom) instances as instance) o + LEFT OUTER JOIN ( + SELECT substr(d.id, 4) id + from ${openaire_db_name}.datasource d + WHERE d.datainfo.deletedbyinference=false) d on o.datasource = d.id; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15.sql new file mode 100644 index 000000000..60b37048b --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step15.sql @@ -0,0 +1,36 @@ +------------------------------------------------------ +------------------------------------------------------ +-- Additional relations +-- +-- Refereed related tables/views +------------------------------------------------------ +------------------------------------------------------ + +CREATE TABLE IF NOT EXISTS ${stats_db_name}.publication_refereed as +select substr(r.id, 4) as id, inst.refereed.classname as refereed +from ${openaire_db_name}.publication r lateral view explode(r.instance) instances as inst +where r.datainfo.deletedbyinference=false; + +CREATE TABLE IF NOT EXISTS ${stats_db_name}.dataset_refereed as +select substr(r.id, 4) as id, inst.refereed.classname as refereed +from ${openaire_db_name}.dataset r lateral view explode(r.instance) instances as inst +where r.datainfo.deletedbyinference=false; + +CREATE TABLE IF NOT EXISTS ${stats_db_name}.software_refereed as +select substr(r.id, 4) as id, inst.refereed.classname as refereed +from ${openaire_db_name}.software r lateral view explode(r.instance) instances as inst +where r.datainfo.deletedbyinference=false; + +CREATE TABLE IF NOT EXISTS ${stats_db_name}.otherresearchproduct_refereed as +select substr(r.id, 4) as id, inst.refereed.classname as refereed +from ${openaire_db_name}.otherresearchproduct r lateral view explode(r.instance) instances as inst +where r.datainfo.deletedbyinference=false; + +CREATE VIEW IF NOT EXISTS ${stats_db_name}.result_refereed as +select * from ${stats_db_name}.publication_refereed +union all +select * from ${stats_db_name}.dataset_refereed +union all +select * from ${stats_db_name}.software_refereed +union all +select * from ${stats_db_name}.otherresearchproduct_refereed; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16.sql new file mode 100644 index 000000000..33849b960 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16.sql @@ -0,0 +1,80 @@ +---------------------------------------------------- +-- Shortcuts for various definitions in stats db --- +---------------------------------------------------- + +-- Peer reviewed: +-- Results that have been collected from Crossref +create table ${stats_db_name}.result_peerreviewed as +with peer_reviewed as ( + select distinct r.id as id + from ${stats_db_name}.result r + join ${stats_db_name}.result_sources rs on rs.id=r.id + join ${stats_db_name}.datasource d on d.id=rs.datasource + where d.name='Crossref') +select distinct peer_reviewed.id as id, true as peer_reviewed +from peer_reviewed +union all +select distinct r.id as id, false as peer_reviewed +from ${stats_db_name}.result r +left outer join peer_reviewed pr on pr.id=r.id +where pr.id is null; + +-- Green OA: +-- OA results that are hosted by an Institutional repository and have NOT been harvested from a DOAJ journal. +create table ${stats_db_name}.result_greenoa as +with result_green as ( + select distinct r.id as id + from ${stats_db_name}.result r + join ${stats_db_name}.result_datasources rd on rd.id=r.id + join ${stats_db_name}.datasource d on d.id=rd.datasource + left outer join ( + select rd.id from ${stats_db_name}.result_datasources rd + join ${stats_db_name}.datasource d on rd.datasource=d.id + join ${stats_db_name}.datasource_sources sds on sds.id=d.id + join ${stats_db_name}.datasource sd on sd.id=sds.datasource + where sd.name='DOAJ-ARTICLES' + ) as doaj on doaj.id=r.id + where r.bestlicence in ('Open Access', 'Open Source') and d.type='Institutional Repository' and doaj.id is null) +select distinct result_green.id, true as green +from result_green +union all +select distinct r.id as id, false as green +from ${stats_db_name}.result r +left outer join result_green rg on rg.id=r.id +where rg.id is null; + +-- GOLD OA: +-- OA results that have been harvested from a DOAJ journal. +create table ${stats_db_name}.result_gold as +with result_gold as ( + select distinct r.id as id + from ${stats_db_name}.result r + join ${stats_db_name}.result_datasources rd on rd.id=r.id + join ${stats_db_name}.datasource d on d.id=rd.datasource + join ${stats_db_name}.datasource_sources sds on sds.id=d.id + join ${stats_db_name}.datasource sd on sd.id=sds.datasource + where r.type='publication' and r.bestlicence='Open Access' and sd.name='DOAJ-Articles') +select distinct result_gold.id, true as gold +from result_gold +union all +select distinct r.id, false as gold +from ${stats_db_name}.result r +where r.id not in (select id from result_gold); + +-- shortcut result-country through the organization affiliation +create table ${stats_db_name}.result_affiliated_country as +select r.id as id, o.country as country +from ${stats_db_name}.result r +join ${stats_db_name}.result_organization ro on ro.id=r.id +join ${stats_db_name}.organization o on o.id=ro.organization +where o.country is not null and o.country!=''; + +-- shortcut result-country through datasource of deposition +create table ${stats_db_name}.result_deposited_country as +select r.id as id, o.country as country +from ${stats_db_name}.result r +join ${stats_db_name}.result_datasources rd on rd.id=r.id +join ${stats_db_name}.datasource d on d.id=rd.datasource +join ${stats_db_name}.datasource_organizations dor on dor.id=d.id +join ${stats_db_name}.organization o on o.id=dor.organization +where o.country is not null and o.country!=''; \ No newline at end of file diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_5.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_5.sql new file mode 100644 index 000000000..f737c1ea6 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_5.sql @@ -0,0 +1,55 @@ +-- replace the creation of the result view to include the boolean fields from the previous tables (green, gold, +-- peer reviewed) +drop table if exists ${stats_db_name}.result_tmp; +CREATE TABLE ${stats_db_name}.result_tmp ( + id STRING, + title STRING, + publisher STRING, + journal STRING, + `date` STRING, + `year` INT, + bestlicence STRING, + access_mode STRING, + embargo_end_date STRING, + delayed BOOLEAN, + authors INT, + source STRING, + abstract BOOLEAN, + type STRING , + peer_reviewed BOOLEAN, + green BOOLEAN, + gold BOOLEAN) +clustered by (id) into 100 buckets stored as orc tblproperties('transactional'='true'); + +insert into ${stats_db_name}.result_tmp +select r.id, r.title, r.publisher, r.journal, r.`date`, date_format(r.`date`, 'yyyy'), r.bestlicence, r.bestlicence, r.embargo_end_date, r.delayed, r.authors, r.source, r.abstract, r.type, pr.peer_reviewed, green.green, gold.gold +FROM ${stats_db_name}.publication r +LEFT OUTER JOIN ${stats_db_name}.result_peerreviewed pr on pr.id=r.id +LEFT OUTER JOIN ${stats_db_name}.result_greenoa green on green.id=r.id +LEFT OUTER JOIN ${stats_db_name}.result_gold gold on gold.id=r.id; + +insert into ${stats_db_name}.result_tmp +select r.id, r.title, r.publisher, r.journal, r.`date`, date_format(r.`date`, 'yyyy'), r.bestlicence, r.bestlicence, r.embargo_end_date, r.delayed, r.authors, r.source, r.abstract, r.type, pr.peer_reviewed, green.green, gold.gold +FROM ${stats_db_name}.dataset r +LEFT OUTER JOIN ${stats_db_name}.result_peerreviewed pr on pr.id=r.id +LEFT OUTER JOIN ${stats_db_name}.result_greenoa green on green.id=r.id +LEFT OUTER JOIN ${stats_db_name}.result_gold gold on gold.id=r.id; + +insert into ${stats_db_name}.result_tmp +select r.id, r.title, r.publisher, r.journal, r.`date`, date_format(r.`date`, 'yyyy'), r.bestlicence, r.bestlicence, r.embargo_end_date, r.delayed, r.authors, r.source, r.abstract, r.type, pr.peer_reviewed, green.green, gold.gold +FROM ${stats_db_name}.software r +LEFT OUTER JOIN ${stats_db_name}.result_peerreviewed pr on pr.id=r.id +LEFT OUTER JOIN ${stats_db_name}.result_greenoa green on green.id=r.id +LEFT OUTER JOIN ${stats_db_name}.result_gold gold on gold.id=r.id; + +insert into ${stats_db_name}.result_tmp +select r.id, r.title, r.publisher, r.journal, r.`date`, date_format(r.`date`, 'yyyy'), r.bestlicence, r.bestlicence, r.embargo_end_date, r.delayed, r.authors, r.source, r.abstract, r.type, pr.peer_reviewed, green.green, gold.gold +FROM ${stats_db_name}.otherresearchproduct r +LEFT OUTER JOIN ${stats_db_name}.result_peerreviewed pr on pr.id=r.id +LEFT OUTER JOIN ${stats_db_name}.result_greenoa green on green.id=r.id +LEFT OUTER JOIN ${stats_db_name}.result_gold gold on gold.id=r.id; + +drop table if exists ${stats_db_name}.result; +drop view if exists ${stats_db_name}.result; +create table ${stats_db_name}.result stored as parquet as select * from ${stats_db_name}.result_tmp; +drop table ${stats_db_name}.result_tmp; \ No newline at end of file diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_6.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_6.sql new file mode 100644 index 000000000..ced7bbc11 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16_6.sql @@ -0,0 +1,32 @@ +------------------------------------------- +--- Extra tables, mostly used by indicators + +create table ${stats_db_name}.result_projectcount as +select r.id, count(distinct p.id) as count +from ${stats_db_name}.result r +left outer join ${stats_db_name}.result_projects rp on rp.id=r.id +left outer join ${stats_db_name}.project p on p.id=rp.project +group by r.id; + +create table ${stats_db_name}.result_fundercount as +select r.id, count(distinct p.funder) as count +from ${stats_db_name}.result r +left outer join ${stats_db_name}.result_projects rp on rp.id=r.id +left outer join ${stats_db_name}.project p on p.id=rp.project +group by r.id; + +create table ${stats_db_name}.project_resultcount as +with rcount as ( + select p.id as pid, count(distinct r.id) as `count`, r.type as type + from ${stats_db_name}.project p + left outer join ${stats_db_name}.result_projects rp on rp.project=p.id + left outer join ${stats_db_name}.result r on r.id=rp.id + group by r.type, p.id ) +select rcount.pid, sum(case when rcount.type='publication' then rcount.count else 0 end) as publications, + sum(case when rcount.type='dataset' then rcount.count else 0 end) as datasets, + sum(case when rcount.type='software' then rcount.count else 0 end) as software, + sum(case when rcount.type='other' then rcount.count else 0 end) as other +from rcount +group by rcount.pid; + +create view ${stats_db_name}.rndexpenditure as select * from stats_ext.rndexpediture \ No newline at end of file diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step17.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step17.sql new file mode 100644 index 000000000..5c102d014 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step17.sql @@ -0,0 +1,207 @@ +------------------------------------------------------ +------------------------------------------------------ +-- Shadow schema table exchange +------------------------------------------------------ +------------------------------------------------------ + +-- Dropping old views +DROP VIEW IF EXISTS ${stats_db_shadow_name}.category; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.concept; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.context; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.country; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.countrygdp; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.creation_date; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_citations; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_classifications; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_concepts; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_datasources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_languages; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_licenses; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_oids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_pids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_refereed; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_sources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.dataset_topics; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.datasource; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.datasource_languages; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.datasource_oids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.datasource_organizations; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.datasource_results; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.datasource_sources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.funder; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.fundref; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.numbers_country; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.organization; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.organization_datasources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.organization_pids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.organization_projects; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.organization_sources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_citations; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_classifications; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_concepts; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_datasources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_languages; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_licenses; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_oids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_pids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_refereed; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_sources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.otherresearchproduct_topics; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.project; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.project_oids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.project_organizations; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.project_results; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.project_resultcount; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.project_results_publication; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_citations; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_classifications; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_concepts; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_datasources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_languages; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_licenses; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_oids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_pids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_refereed; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_sources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.publication_topics; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_affiliated_country; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_citations; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_classifications; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_concepts; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_datasources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_deposited_country; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_fundercount; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_gold; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_greenoa; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_languages; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_licenses; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_oids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_organization; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_peerreviewed; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_pids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_projectcount; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_projects; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_refereed; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_sources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.result_topics; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.rndexpediture; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.roarmap; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_citations; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_classifications; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_concepts; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_datasources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_languages; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_licenses; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_oids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_pids; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_refereed; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_sources; +DROP VIEW IF EXISTS ${stats_db_shadow_name}.software_topics; + + +-- Creating the shadow database, in case it doesn't exist +CREATE database IF NOT EXISTS ${stats_db_shadow_name}; + +-- Creating new views +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.category AS SELECT * FROM ${stats_db_name}.category; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.concept AS SELECT * FROM ${stats_db_name}.concept; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.context AS SELECT * FROM ${stats_db_name}.context; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.country AS SELECT * FROM ${stats_db_name}.country; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.countrygdp AS SELECT * FROM ${stats_db_name}.countrygdp; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.creation_date AS SELECT * FROM ${stats_db_name}.creation_date; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset AS SELECT * FROM ${stats_db_name}.dataset; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_citations AS SELECT * FROM ${stats_db_name}.dataset_citations; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_classifications AS SELECT * FROM ${stats_db_name}.dataset_classifications; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_concepts AS SELECT * FROM ${stats_db_name}.dataset_concepts; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_datasources AS SELECT * FROM ${stats_db_name}.dataset_datasources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_languages AS SELECT * FROM ${stats_db_name}.dataset_languages; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_licenses AS SELECT * FROM ${stats_db_name}.dataset_licenses; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_oids AS SELECT * FROM ${stats_db_name}.dataset_oids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_pids AS SELECT * FROM ${stats_db_name}.dataset_pids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_refereed AS SELECT * FROM ${stats_db_name}.dataset_refereed; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_sources AS SELECT * FROM ${stats_db_name}.dataset_sources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.dataset_topics AS SELECT * FROM ${stats_db_name}.dataset_topics; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.datasource AS SELECT * FROM ${stats_db_name}.datasource; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.datasource_languages AS SELECT * FROM ${stats_db_name}.datasource_languages; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.datasource_oids AS SELECT * FROM ${stats_db_name}.datasource_oids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.datasource_organizations AS SELECT * FROM ${stats_db_name}.datasource_organizations; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.datasource_results AS SELECT * FROM ${stats_db_name}.datasource_results; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.datasource_sources AS SELECT * FROM ${stats_db_name}.datasource_sources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.funder AS SELECT * FROM ${stats_db_name}.funder; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.fundref AS SELECT * FROM ${stats_db_name}.fundref; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.numbers_country AS SELECT * FROM ${stats_db_name}.numbers_country; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.organization AS SELECT * FROM ${stats_db_name}.organization; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.organization_datasources AS SELECT * FROM ${stats_db_name}.organization_datasources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.organization_pids AS SELECT * FROM ${stats_db_name}.organization_pids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.organization_projects AS SELECT * FROM ${stats_db_name}.organization_projects; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.organization_sources AS SELECT * FROM ${stats_db_name}.organization_sources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct AS SELECT * FROM ${stats_db_name}.otherresearchproduct; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_citations AS SELECT * FROM ${stats_db_name}.otherresearchproduct_citations; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_classifications AS SELECT * FROM ${stats_db_name}.otherresearchproduct_classifications; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_concepts AS SELECT * FROM ${stats_db_name}.otherresearchproduct_concepts; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_datasources AS SELECT * FROM ${stats_db_name}.otherresearchproduct_datasources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_languages AS SELECT * FROM ${stats_db_name}.otherresearchproduct_languages; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_licenses AS SELECT * FROM ${stats_db_name}.otherresearchproduct_licenses; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_oids AS SELECT * FROM ${stats_db_name}.otherresearchproduct_oids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_pids AS SELECT * FROM ${stats_db_name}.otherresearchproduct_pids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_refereed AS SELECT * FROM ${stats_db_name}.otherresearchproduct_refereed; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_sources AS SELECT * FROM ${stats_db_name}.otherresearchproduct_sources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.otherresearchproduct_topics AS SELECT * FROM ${stats_db_name}.otherresearchproduct_topics; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.project AS SELECT * FROM ${stats_db_name}.project; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.project_oids AS SELECT * FROM ${stats_db_name}.project_oids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.project_organizations AS SELECT * FROM ${stats_db_name}.project_organizations; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.project_results AS SELECT * FROM ${stats_db_name}.project_results; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.project_resultcount AS SELECT * FROM ${stats_db_name}.project_resultcount; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.project_results_publication AS SELECT * FROM ${stats_db_name}.project_results_publication; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication AS SELECT * FROM ${stats_db_name}.publication; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_citations AS SELECT * FROM ${stats_db_name}.publication_citations; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_classifications AS SELECT * FROM ${stats_db_name}.publication_classifications; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_concepts AS SELECT * FROM ${stats_db_name}.publication_concepts; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_datasources AS SELECT * FROM ${stats_db_name}.publication_datasources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_languages AS SELECT * FROM ${stats_db_name}.publication_languages; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_licenses AS SELECT * FROM ${stats_db_name}.publication_licenses; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_oids AS SELECT * FROM ${stats_db_name}.publication_oids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_pids AS SELECT * FROM ${stats_db_name}.publication_pids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_refereed AS SELECT * FROM ${stats_db_name}.publication_refereed; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_sources AS SELECT * FROM ${stats_db_name}.publication_sources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.publication_topics AS SELECT * FROM ${stats_db_name}.publication_topics; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result AS SELECT * FROM ${stats_db_name}.result; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_affiliated_country AS SELECT * FROM ${stats_db_name}.result_affiliated_country; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_citations AS SELECT * FROM ${stats_db_name}.result_citations; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_classifications AS SELECT * FROM ${stats_db_name}.result_classifications; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_concepts AS SELECT * FROM ${stats_db_name}.result_concepts; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_datasources AS SELECT * FROM ${stats_db_name}.result_datasources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_deposited_country AS SELECT * FROM ${stats_db_name}.result_deposited_country; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_fundercount AS SELECT * FROM ${stats_db_name}.result_fundercount; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_gold AS SELECT * FROM ${stats_db_name}.result_gold; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_greenoa AS SELECT * FROM ${stats_db_name}.result_greenoa; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_languages AS SELECT * FROM ${stats_db_name}.result_languages; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_licenses AS SELECT * FROM ${stats_db_name}.result_licenses; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_oids AS SELECT * FROM ${stats_db_name}.result_oids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_organization AS SELECT * FROM ${stats_db_name}.result_organization; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_peerreviewed AS SELECT * FROM ${stats_db_name}.result_peerreviewed; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_pids AS SELECT * FROM ${stats_db_name}.result_pids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_projectcount AS SELECT * FROM ${stats_db_name}.result_projectcount; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_projects AS SELECT * FROM ${stats_db_name}.result_projects; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_refereed AS SELECT * FROM ${stats_db_name}.result_refereed; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_sources AS SELECT * FROM ${stats_db_name}.result_sources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.result_topics AS SELECT * FROM ${stats_db_name}.result_topics; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.rndexpediture AS SELECT * FROM ${stats_db_name}.rndexpediture; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.roarmap AS SELECT * FROM ${stats_db_name}.roarmap; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software AS SELECT * FROM ${stats_db_name}.software; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_citations AS SELECT * FROM ${stats_db_name}.software_citations; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_classifications AS SELECT * FROM ${stats_db_name}.software_classifications; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_concepts AS SELECT * FROM ${stats_db_name}.software_concepts; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_datasources AS SELECT * FROM ${stats_db_name}.software_datasources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_languages AS SELECT * FROM ${stats_db_name}.software_languages; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_licenses AS SELECT * FROM ${stats_db_name}.software_licenses; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_oids AS SELECT * FROM ${stats_db_name}.software_oids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_pids AS SELECT * FROM ${stats_db_name}.software_pids; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_refereed AS SELECT * FROM ${stats_db_name}.software_refereed; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_sources AS SELECT * FROM ${stats_db_name}.software_sources; +CREATE VIEW IF NOT EXISTS ${stats_db_shadow_name}.software_topics AS SELECT * FROM ${stats_db_name}.software_topics; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step18.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step18.sql new file mode 100644 index 000000000..34e48a18a --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step18.sql @@ -0,0 +1,8 @@ +------------------------------------------------------ +------------------------------------------------------ +-- Impala table statistics - Needed to make the tables +-- visible for impala +------------------------------------------------------ +------------------------------------------------------ + +INVALIDATE METADATA ${stats_db_name}; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step19.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step19.sql new file mode 100644 index 000000000..34e48a18a --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step19.sql @@ -0,0 +1,8 @@ +------------------------------------------------------ +------------------------------------------------------ +-- Impala table statistics - Needed to make the tables +-- visible for impala +------------------------------------------------------ +------------------------------------------------------ + +INVALIDATE METADATA ${stats_db_name}; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step2.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step2.sql new file mode 100644 index 000000000..ba0db25be --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step2.sql @@ -0,0 +1,44 @@ +-------------------------------------------------------------- +-------------------------------------------------------------- +-- Publication table/view and Publication related tables/views +-------------------------------------------------------------- +-------------------------------------------------------------- + +-- Publication temporary table +DROP TABLE IF EXISTS ${stats_db_name}.publication_tmp; + +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 +from ${openaire_db_name}.publication p +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_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}.publication 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}.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_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; + +-- 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; \ No newline at end of file diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step3.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step3.sql new file mode 100644 index 000000000..f69715a31 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step3.sql @@ -0,0 +1,36 @@ +------------------------------------------------------ +------------------------------------------------------ +-- Dataset table/view and Dataset related tables/views +------------------------------------------------------ +------------------------------------------------------ + +-- Dataset temporary table supporting updates +DROP TABLE IF EXISTS ${stats_db_name}.dataset_tmp; +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, + CASE WHEN SIZE(d.description) > 0 THEN TRUE ELSE FALSE end AS abstract, +'dataset' AS type +FROM ${openaire_db_name}.dataset d +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_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_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_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_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; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step4.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step4.sql new file mode 100644 index 000000000..2c4a625e1 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step4.sql @@ -0,0 +1,36 @@ +-------------------------------------------------------- +-------------------------------------------------------- +-- Software table/view and Software related tables/views +-------------------------------------------------------- +-------------------------------------------------------- + +-- Software temporary table supporting updates +DROP TABLE IF EXISTS ${stats_db_name}.software_tmp; +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, + CASE WHEN SIZE(s.description) > 0 THEN TRUE ELSE FALSE END AS abstract, +'software' as type +from ${openaire_db_name}.software s +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_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_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_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_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; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step5.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step5.sql new file mode 100644 index 000000000..1fa5df8cb --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step5.sql @@ -0,0 +1,36 @@ +-------------------------------------------------------------------------------- +-------------------------------------------------------------------------------- +-- Otherresearchproduct table/view and Otherresearchproduct related tables/views +-------------------------------------------------------------------------------- +-------------------------------------------------------------------------------- + +-- Otherresearchproduct temporary table supporting updates +DROP TABLE IF EXISTS ${stats_db_name}.otherresearchproduct_tmp; +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 +FROM ${openaire_db_name}.otherresearchproduct o +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_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_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_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_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; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step6.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step6.sql new file mode 100644 index 000000000..21a944164 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step6.sql @@ -0,0 +1,30 @@ +-- noinspection SqlNoDataSourceInspectionForFile + +------------------------------------------------------ +------------------------------------------------------ +-- Project table/view and Project related tables/views +------------------------------------------------------ +------------------------------------------------------ +-- Project_oids Table +DROP TABLE IF EXISTS ${stats_db_name}.project_oids; +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; + +-- Project_organizations Table +DROP TABLE IF EXISTS ${stats_db_name}.project_organizations; +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'; + +-- Project_results Table +DROP TABLE IF EXISTS ${stats_db_name}.project_results; +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 table +---------------- +-- Creating and populating temporary Project table +DROP TABLE IF EXISTS ${stats_db_name}.project_tmp; +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; + +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 +from ${openaire_db_name}.project p lateral view explode(p.fundingtree.value) fundingtree as fund diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql new file mode 100644 index 000000000..7acabf1dd --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step7.sql @@ -0,0 +1,31 @@ +---------------------------------------------------- +---------------------------------------------------- +-- Result table/view and Result related tables/views +---------------------------------------------------- +---------------------------------------------------- + +-- 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; + +-- 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_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_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_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_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; + +DROP TABLE IF EXISTS ${stats_db_name}.result_organization; +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'; + +DROP TABLE IF EXISTS ${stats_db_name}.result_projects; +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; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step8.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step8.sql new file mode 100644 index 000000000..4e13b3dd8 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step8.sql @@ -0,0 +1,58 @@ +-- noinspection SqlNoDataSourceInspectionForFile + +------------------------------------------------------------ +------------------------------------------------------------ +-- 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'); + +-- 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 +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; + +-- 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'); +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'; + +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; + +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; + +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'; + +-- 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 OR REPLACE VIEW ${stats_db_name}.datasource_results AS SELECT datasource AS id, id AS result FROM ${stats_db_name}.result_datasources; + + diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step9.sql b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step9.sql new file mode 100644 index 000000000..a918e4de4 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step9.sql @@ -0,0 +1,12 @@ +---------------------------------------------------------------- +---------------------------------------------------------------- +-- 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 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; diff --git a/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml new file mode 100644 index 000000000..324e6f9a1 --- /dev/null +++ b/dhp-workflows/dhp-stats-promote/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/workflow.xml @@ -0,0 +1,302 @@ + + + + stats_db_name + the target stats database name + + + stats_db_shadow_name + the name of the shadow schema + + + stats_db_production_name + the name of the production schema + + + hive_metastore_uris + hive server metastore URIs + + + hive_jdbc_url + hive server jdbc url + + + hive_timeout + the time period, in seconds, after which Hive fails a transaction if a Hive client has not sent a hearbeat. The default value is 300 seconds. + + + + + ${jobTracker} + ${nameNode} + + + hive.metastore.uris + ${hive_metastore_uris} + + + hive.txn.timeout + ${hive_timeout} + + + + + + + + Action failed, error message[${wf:errorMessage(wf:lastErrorNode())}] + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + external_stats_db_name=${external_stats_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + external_stats_db_name=${external_stats_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + openaire_db_name=${openaire_db_name} + + + + + + + + ${hive_jdbc_url} + + stats_db_name=${stats_db_name} + stats_db_shadow_name=${stats_db_shadow_name} + + + + + + + + ${jobTracker} + ${nameNode} + impala-shell.sh + ${stats_db_name} + step18.sql + ${wf:appPath()}/scripts/step18.sql + impala-shell.sh + + + + + + + + ${jobTracker} + ${nameNode} + impala-shell.sh + ${stats_db_shadow_name} + step19.sql + ${wf:appPath()}/scripts/step19.sql + impala-shell.sh + + + + + + + + ${jobTracker} + ${nameNode} + updateCache.sh + ${stats_tool_api_url} + updateCache.sh + + + + + + +