sabeel
/
dnet-hadoop
forked from D-Net/dnet-hadoop
1
0
Fork 0
Code Pull Requests Activity

resolving conflicts on step16-createIndicatorsTables.sql

This commit is contained in:
Claudio Atzori 2024-03-26 12:17:52 +01:00
parent f6601ea7d1
commit d70793847d
1 changed files with 243 additions and 405 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

648
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16-createIndicatorsTables.sql
View File

@ -1,390 +1,255 @@
-- Sprint 1 ----
drop table if exists ${stats_db_name}.indi_pub_green_oa purge; /*EOS*/
--create table if not exists ${stats_db_name}.indi_pub_green_oa stored as parquet as
--select distinct p.id, coalesce(green_oa, 0) as green_oa
--from ${stats_db_name}.publication p
-- left outer join (
-- select p.id, 1 as green_oa
-- from ${stats_db_name}.publication p
-- join ${stats_db_name}.result_instance ri on ri.id = p.id
-- join ${stats_db_name}.datasource on datasource.id = ri.hostedby
-- where datasource.type like '%Repository%'
-- and (ri.accessright = 'Open Access'
-- or ri.accessright = 'Embargo' or ri.accessright = 'Open Source')) tmp
-- on p.id= tmp.id;
create table if not exists ${stats_db_name}.indi_pub_green_oa stored as parquet as
select distinct p.id, coalesce(green_oa, 0) as green_oa
from ${stats_db_name}.publication p
left outer join (
left outer join (
select p.id, 1 as green_oa
from ${stats_db_name}.publication p
join ${stats_db_name}.result_instance ri on ri.id = p.id
join ${stats_db_name}.datasource on datasource.id = ri.hostedby
where datasource.type like '%Repository%'
and (ri.accessright = 'Open Access'
or ri.accessright = 'Embargo' or ri.accessright = 'Open Source') and datasource.name!='Other') tmp
on p.id= tmp.id; /*EOS*/
join ${stats_db_name}.result_instance ri on ri.id = p.id
join ${stats_db_name}.datasource on datasource.id = ri.hostedby
where datasource.type like '%Repository%' and (ri.accessright = 'Open Access' or ri.accessright = 'Embargo' or ri.accessright = 'Open Source') and datasource.name!='Other') tmp on p.id= tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_grey_lit purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_grey_lit stored as parquet as
select distinct p.id, coalesce(grey_lit, 0) as grey_lit
from ${stats_db_name}.publication p
left outer join (
left outer join (
select p.id, 1 as grey_lit
from ${stats_db_name}.publication p
join ${stats_db_name}.result_classifications rt on rt.id = p.id
where rt.type not in ('Article','Part of book or chapter of book','Book','Doctoral thesis','Master thesis','Data Paper', 'Thesis', 'Bachelor thesis', 'Conference object') and
not exists (select 1 from ${stats_db_name}.result_classifications rc where type ='Other literature type'
and rc.id=p.id)) tmp on p.id=tmp.id; /*EOS*/
join ${stats_db_name}.result_classifications rt on rt.id = p.id
where rt.type not in ('Article','Part of book or chapter of book','Book','Doctoral thesis','Master thesis','Data Paper', 'Thesis', 'Bachelor thesis', 'Conference object')
and not exists (select 1 from ${stats_db_name}.result_classifications rc where type ='Other literature type' and rc.id=p.id)) tmp on p.id=tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_doi_from_crossref purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_doi_from_crossref stored as parquet as
select distinct p.id, coalesce(doi_from_crossref, 0) as doi_from_crossref
from ${stats_db_name}.publication p
left outer join
(select ri.id, 1 as doi_from_crossref from ${stats_db_name}.result_instance ri
join ${stats_db_name}.datasource d on d.id = ri.collectedfrom
where pidtype='Digital Object Identifier' and d.name ='Crossref') tmp
on tmp.id=p.id; /*EOS*/
left outer join (
select ri.id, 1 as doi_from_crossref from ${stats_db_name}.result_instance ri
join ${stats_db_name}.datasource d on d.id = ri.collectedfrom
where pidtype='Digital Object Identifier' and d.name ='Crossref') tmp on tmp.id=p.id; /*EOS*/
-- Sprint 2 ----
drop table if exists ${stats_db_name}.indi_result_has_cc_licence purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_result_has_cc_licence stored as parquet as
select distinct r.id, (case when lic='' or lic is null then 0 else 1 end) as has_cc_license
from ${stats_db_name}.result r
left outer join (select r.id, license.type as lic from ${stats_db_name}.result r
join ${stats_db_name}.result_licenses as license on license.id = r.id
where lower(license.type) LIKE '%creativecommons.org%' OR lower(license.type) LIKE '%cc-%') tmp
on r.id= tmp.id; /*EOS*/
left outer join (
select r.id, license.type as lic from ${stats_db_name}.result r
join ${stats_db_name}.result_licenses as license on license.id = r.id
where lower(license.type) LIKE '%creativecommons.org%' OR lower(license.type) LIKE '%cc %') tmp on r.id= tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_result_has_cc_licence_url purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_result_has_cc_licence_url stored as parquet as
select distinct r.id, case when lic_host='' or lic_host is null then 0 else 1 end as has_cc_license_url
from ${stats_db_name}.result r
left outer join (select r.id, lower(parse_url(license.type, "HOST")) as lic_host
from ${stats_db_name}.result r
join ${stats_db_name}.result_licenses as license on license.id = r.id
WHERE lower(parse_url(license.type, "HOST")) = "creativecommons.org") tmp
on r.id= tmp.id; /*EOS*/
left outer join (
select r.id, lower(parse_url(license.type, "HOST")) as lic_host
from ${stats_db_name}.result r
join ${stats_db_name}.result_licenses as license on license.id = r.id
WHERE lower(parse_url(license.type, "HOST")) = "creativecommons.org") tmp on r.id= tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_has_abstract purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_has_abstract stored as parquet as
select distinct publication.id, cast(coalesce(abstract, true) as int) has_abstract
from ${stats_db_name}.publication; /*EOS*/
drop table if exists ${stats_db_name}.indi_result_with_orcid purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_result_with_orcid stored as parquet as
select distinct r.id, coalesce(has_orcid, 0) as has_orcid
from ${stats_db_name}.result r
left outer join (select id, 1 as has_orcid from ${stats_db_name}.result_orcid) tmp
on r.id= tmp.id; /*EOS*/
left outer join (
select id, 1 as has_orcid from ${stats_db_name}.result_orcid) tmp on r.id= tmp.id; /*EOS*/
---- Sprint 3 ----
drop table if exists ${stats_db_name}.indi_funded_result_with_fundref purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_funded_result_with_fundref stored as parquet as
select distinct r.result as id, coalesce(fundref, 0) as fundref
from ${stats_db_name}.project_results r
left outer join (select distinct result, 1 as fundref from ${stats_db_name}.project_results
where provenance='Harvested') tmp
on r.result= tmp.result; /*EOS*/
-- create table indi_result_org_collab stored as parquet as
-- select o1.organization org1, o2.organization org2, count(distinct o1.id) as collaborations
-- from result_organization as o1
-- join result_organization as o2 on o1.id=o2.id and o1.organization!=o2.organization
-- group by o1.organization, o2.organization;
--
-- compute stats indi_result_org_collab;
--
create TEMPORARY VIEW tmp AS SELECT ro.organization organization, ro.id, o.name from ${stats_db_name}.result_organization ro
join ${stats_db_name}.organization o on o.id=ro.organization where o.name is not null; /*EOS*/
left outer join (
select distinct result, 1 as fundref from ${stats_db_name}.project_results where provenance='Harvested') tmp on r.result= tmp.result; /*EOS*/
drop table if exists ${stats_db_name}.indi_result_org_collab purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_result_org_collab stored as parquet as
select o1.organization org1, o1.name org1name1, o2.organization org2, o2.name org2name2, count(o1.id) as collaborations
from tmp as o1
join tmp as o2 where o1.id=o2.id and o1.organization!=o2.organization and o1.name!=o2.name
group by o1.organization, o2.organization, o1.name, o2.name; /*EOS*/
DROP VIEW if exists tmp; /*EOS*/
create TEMPORARY VIEW tmp AS
select distinct ro.organization organization, ro.id, o.name, o.country from ${stats_db_name}.result_organization ro
join ${stats_db_name}.organization o on o.id=ro.organization where country <> 'UNKNOWN' and o.name is not null; /*EOS*/
WITH tmp AS (
SELECT ro.organization organization, ro.id, o.name
from ${stats_db_name}.result_organization ro
join ${stats_db_name}.organization o on o.id=ro.organization where o.name is not null)
select o1.organization org1, o1.name org1name1, o2.organization org2, o2.name org2name2, count(o1.id) as collaborations
from tmp as o1
join tmp as o2 where o1.id=o2.id and o1.organization!=o2.organization and o1.name!=o2.name
group by o1.organization, o2.organization, o1.name, o2.name; /*EOS*/
drop table if exists ${stats_db_name}.indi_result_org_country_collab purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_result_org_country_collab stored as parquet as
select o1.organization org1,o1.name org1name1, o2.country country2, count(o1.id) as collaborations
from tmp as o1 join tmp as o2 on o1.id=o2.id
where o1.id=o2.id and o1.country!=o2.country
group by o1.organization, o1.id, o1.name, o2.country; /*EOS*/
drop table if exists tmp purge; /*EOS*/
create TEMPORARY VIEW tmp AS
select o.id organization, o.name, ro.project as project from ${stats_db_name}.organization o
join ${stats_db_name}.organization_projects ro on o.id=ro.id where o.name is not null; /*EOS*/
WITH tmp AS (
select distinct ro.organization organization, ro.id, o.name, o.country
from ${stats_db_name}.result_organization ro
join ${stats_db_name}.organization o on o.id=ro.organization
where country <> 'UNKNOWN' and o.name is not null)
select o1.organization org1,o1.name org1name1, o2.country country2, count(o1.id) as collaborations
from tmp as o1 join tmp as o2 on o1.id=o2.id
where o1.id=o2.id and o1.country!=o2.country
group by o1.organization, o1.id, o1.name, o2.country; /*EOS*/
drop table if exists ${stats_db_name}.indi_project_collab_org purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_project_collab_org stored as parquet as
select o1.organization org1,o1.name orgname1, o2.organization org2, o2.name orgname2, count(distinct o1.project) as collaborations
from tmp as o1
join tmp as o2 on o1.project=o2.project
where o1.organization<>o2.organization and o1.name<>o2.name
group by o1.name,o2.name, o1.organization, o2.organization; /*EOS*/
DROP VIEW if exists tmp; /*EOS*/
create TEMPORARY VIEW tmp AS
select o.id organization, o.name, o.country , ro.project as project from ${stats_db_name}.organization o
join ${stats_db_name}.organization_projects ro on o.id=ro.id
and o.country <> 'UNKNOWN' and o.name is not null; /*EOS*/
WITH tmp AS (
select o.id organization, o.name, ro.project as project
from ${stats_db_name}.organization o
join ${stats_db_name}.organization_projects ro on o.id=ro.id where o.name is not null)
select o1.organization org1,o1.name orgname1, o2.organization org2, o2.name orgname2, count(distinct o1.project) as collaborations
from tmp as o1
join tmp as o2 on o1.project=o2.project
where o1.organization<>o2.organization and o1.name<>o2.name
group by o1.name,o2.name, o1.organization, o2.organization; /*EOS*/
drop table if exists ${stats_db_name}.indi_project_collab_org_country purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_project_collab_org_country stored as parquet as
select o1.organization org1,o1.name org1name, o2.country country2, count(distinct o1.project) as collaborations
from tmp as o1
join tmp as o2 on o1.project=o2.project
where o1.organization<>o2.organization and o1.country<>o2.country
group by o1.organization, o2.country, o1.name; /*EOS*/
DROP VIEW if exists tmp; /*EOS*/
WITH tmp AS (
select o.id organization, o.name, o.country , ro.project as project
from ${stats_db_name}.organization o
join ${stats_db_name}.organization_projects ro on o.id=ro.id and o.country <> 'UNKNOWN' and o.name is not null)
select o1.organization org1,o1.name org1name, o2.country country2, count(distinct o1.project) as collaborations
from tmp as o1
join tmp as o2 on o1.project=o2.project
where o1.organization<>o2.organization and o1.country<>o2.country
group by o1.organization, o2.country, o1.name; /*EOS*/
drop table if exists ${stats_db_name}.indi_funder_country_collab purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_funder_country_collab stored as parquet as
with tmp as (select funder, project, country from ${stats_db_name}.organization_projects op
join ${stats_db_name}.organization o on o.id=op.id
join ${stats_db_name}.project p on p.id=op.project
where country <> 'UNKNOWN')
select f1.funder, f1.country as country1, f2.country as country2, count(distinct f1.project) as collaborations
from tmp as f1
join tmp as f2 on f1.project=f2.project
where f1.country<>f2.country
group by f1.funder, f2.country, f1.country; /*EOS*/
create TEMPORARY VIEW tmp AS
select distinct country, ro.id as result from ${stats_db_name}.organization o
join ${stats_db_name}.result_organization ro on o.id=ro.organization
where country <> 'UNKNOWN' and o.name is not null; /*EOS*/
select f1.funder, f1.country as country1, f2.country as country2, count(distinct f1.project) as collaborations
from tmp as f1
join tmp as f2 on f1.project=f2.project
where f1.country<>f2.country
group by f1.funder, f2.country, f1.country; /*EOS*/
drop table if exists ${stats_db_name}.indi_result_country_collab purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_result_country_collab stored as parquet as
select o1.country country1, o2.country country2, count(o1.result) as collaborations
from tmp as o1
join tmp as o2 on o1.result=o2.result
where o1.country<>o2.country
group by o1.country, o2.country; /*EOS*/
WITH tmp AS (
select distinct country, ro.id as result from ${stats_db_name}.organization o
join ${stats_db_name}.result_organization ro on o.id=ro.organization
where country <> 'UNKNOWN' and o.name is not null)
select o1.country country1, o2.country country2, count(o1.result) as collaborations
from tmp as o1
join tmp as o2 on o1.result=o2.result
where o1.country<>o2.country
group by o1.country, o2.country; /*EOS*/
DROP VIEW if exists tmp; /*EOS*/
---- Sprint 4 ----
drop table if exists ${stats_db_name}.indi_pub_diamond purge; /*EOS*/
--create table if not exists ${stats_db_name}.indi_pub_diamond stored as parquet as
--select distinct pd.id, coalesce(in_diamond_journal, 0) as in_diamond_journal
--from ${stats_db_name}.publication_datasources pd
-- left outer join (
-- select pd.id, 1 as in_diamond_journal from ${stats_db_name}.publication_datasources pd
-- join ${stats_db_name}.datasource d on d.id=pd.datasource
-- join STATS_EXT.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_online=d.issn_online)
-- and (ps.journal_is_in_doaj=true or ps.journal_is_oa=true) and ps.has_apc=false) tmp
-- on pd.id=tmp.id;
create table if not exists ${stats_db_name}.indi_pub_diamond stored as parquet as
select distinct pd.id, coalesce(in_diamond_journal, 0) as in_diamond_journal
from ${stats_db_name}.publication_datasources pd
left outer join (select pd.id, 1 as in_diamond_journal from ${stats_db_name}.publication_datasources pd
join ${stats_db_name}.datasource d on d.id=pd.datasource
join STATS_EXT.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_online=d.issn_online)
and (ps.journal_is_in_doaj=true or ps.journal_is_oa=true) and ps.has_apc=false) tmp
on pd.id=tmp.id; /*EOS*/
select distinct pd.id, coalesce(in_diamond_journal, 0) as in_diamond_journal
from ${stats_db_name}.publication_datasources pd
left outer join (
select pd.id, 1 as in_diamond_journal
from ${stats_db_name}.publication_datasources pd
join ${stats_db_name}.datasource d on d.id=pd.datasource
join STATS_EXT.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_online=d.issn_online)
and (ps.journal_is_in_doaj=true or ps.journal_is_oa=true) and ps.has_apc=false) tmp on pd.id=tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_in_transformative purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_in_transformative stored as parquet as
select distinct pd.id, coalesce(is_transformative, 0) as is_transformative
from ${stats_db_name}.publication pd
left outer join (
select pd.id, 1 as is_transformative from ${stats_db_name}.publication_datasources pd
join ${stats_db_name}.datasource d on d.id=pd.datasource
join STATS_EXT.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_online=d.issn_online)
and ps.is_transformative_journal=true) tmp
on pd.id=tmp.id; /*EOS*/
select distinct pd.id, coalesce(is_transformative, 0) as is_transformative
from ${stats_db_name}.publication pd
left outer join (
select pd.id, 1 as is_transformative
from ${stats_db_name}.publication_datasources pd
join ${stats_db_name}.datasource d on d.id=pd.datasource
join STATS_EXT.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_online=d.issn_online)
and ps.is_transformative_journal=true) tmp on pd.id=tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_closed_other_open purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_closed_other_open stored as parquet as
select distinct ri.id, coalesce(pub_closed_other_open, 0) as pub_closed_other_open from ${stats_db_name}.result_instance ri
left outer join
(select ri.id, 1 as pub_closed_other_open from ${stats_db_name}.result_instance ri
join ${stats_db_name}.publication p on p.id=ri.id
join ${stats_db_name}.datasource d on ri.hostedby=d.id
where d.type like '%Journal%' and ri.accessright='Closed Access' and
(p.bestlicence='Open Access' or p.bestlicence='Open Source')) tmp
on tmp.id=ri.id; /*EOS*/
select distinct ri.id, coalesce(pub_closed_other_open, 0) as pub_closed_other_open
from ${stats_db_name}.result_instance ri
left outer join (
select ri.id, 1 as pub_closed_other_open
from ${stats_db_name}.result_instance ri
join ${stats_db_name}.publication p on p.id=ri.id
join ${stats_db_name}.datasource d on ri.hostedby=d.id
where d.type like '%Journal%' and ri.accessright='Closed Access' and
(p.bestlicence='Open Access' or p.bestlicence='Open Source')) tmp on tmp.id=ri.id; /*EOS*/
---- Sprint 5 ----
drop table if exists ${stats_db_name}.indi_result_no_of_copies purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_result_no_of_copies stored as parquet as
select id, count(id) as number_of_copies from ${stats_db_name}.result_instance group by id; /*EOS*/
select id, count(id) as number_of_copies
from ${stats_db_name}.result_instance
group by id; /*EOS*/
---- Sprint 6 ----
drop table if exists ${stats_db_name}.indi_pub_downloads purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_downloads stored as parquet as
SELECT result_id, sum(downloads) no_downloads from openaire_prod_usage_stats.usage_stats
join ${stats_db_name}.publication on result_id=id
where downloads>0
GROUP BY result_id
order by no_downloads desc; /*EOS*/
--ANALYZE TABLE ${stats_db_name}.indi_pub_downloads COMPUTE STATISTICS;
SELECT result_id, sum(downloads) no_downloads
from openaire_prod_usage_stats.usage_stats
join ${stats_db_name}.publication on result_id=id
where downloads>0
GROUP BY result_id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_downloads_datasource purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_downloads_datasource stored as parquet as
SELECT result_id, repository_id, sum(downloads) no_downloads from openaire_prod_usage_stats.usage_stats
join ${stats_db_name}.publication on result_id=id
where downloads>0
GROUP BY result_id, repository_id
order by result_id; /*EOS*/
SELECT result_id, repository_id, sum(downloads) no_downloads
from openaire_prod_usage_stats.usage_stats
join ${stats_db_name}.publication on result_id=id
where downloads>0
GROUP BY result_id, repository_id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_downloads_year purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_downloads_year stored as parquet as
SELECT result_id, cast(substring(us.`date`, 1,4) as int) as `year`, sum(downloads) no_downloads
from openaire_prod_usage_stats.usage_stats us
join ${stats_db_name}.publication on result_id=id where downloads>0
GROUP BY result_id, substring(us.`date`, 1,4); /*EOS*/
SELECT result_id, cast(substring(us.`date`, 1,4) as int) as `year`, sum(downloads) no_downloads
from openaire_prod_usage_stats.usage_stats us
join ${stats_db_name}.publication on result_id=id where downloads>0
GROUP BY result_id, substring(us.`date`, 1,4); /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_downloads_datasource_year purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_downloads_datasource_year stored as parquet as
SELECT result_id, cast(substring(us.`date`, 1,4) as int) as `year`, repository_id, sum(downloads) no_downloads from openaire_prod_usage_stats.usage_stats us
join ${stats_db_name}.publication on result_id=id
where downloads>0
GROUP BY result_id, repository_id, substring(us.`date`, 1,4); /*EOS*/
SELECT result_id, cast(substring(us.`date`, 1,4) as int) as `year`, repository_id, sum(downloads) no_downloads
from openaire_prod_usage_stats.usage_stats us
join ${stats_db_name}.publication on result_id=id
where downloads>0
GROUP BY result_id, repository_id, substring(us.`date`, 1,4); /*EOS*/
---- Sprint 7 ----
drop table if exists ${stats_db_name}.indi_pub_gold_oa purge; /*EOS*/
--create table if not exists ${stats_db_name}.indi_pub_gold_oa stored as parquet as
-- WITH gold_oa AS ( SELECT
-- issn_l,
-- journal_is_in_doaj,
-- journal_is_oa,
-- issn_1 as issn
-- FROM
-- STATS_EXT.oa_journals
-- WHERE
-- issn_1 != ""
-- UNION
-- ALL SELECT
-- issn_l,
-- journal_is_in_doaj,
-- journal_is_oa,
-- issn_2 as issn
-- FROM
-- STATS_EXT.oa_journals
-- WHERE
-- issn_2 != "" ), issn AS ( SELECT
-- *
-- FROM
--( SELECT
-- id,
-- issn_printed as issn
-- FROM
-- ${stats_db_name}.datasource
-- WHERE
-- issn_printed IS NOT NULL
-- UNION ALL
-- SELECT
-- id,
-- issn_online as issn
-- FROM
-- ${stats_db_name}.datasource
-- WHERE
-- issn_online IS NOT NULL or id like '%doajarticles%') as issn
-- WHERE
-- LENGTH(issn) > 7)
--SELECT
-- DISTINCT pd.id, coalesce(is_gold, 0) as is_gold
--FROM
-- ${stats_db_name}.publication_datasources pd
-- left outer join(
-- select pd.id, 1 as is_gold FROM ${stats_db_name}.publication_datasources pd
-- JOIN issn on issn.id=pd.datasource
-- JOIN gold_oa on issn.issn = gold_oa.issn) tmp
-- on pd.id=tmp.id;
--create table if not exists ${stats_db_name}.indi_pub_gold_oa stored as parquet as
--with gold_oa as (
--SELECT issn,issn_l from stats_ext.issn_gold_oa_dataset_v5),
--issn AS (SELECT * FROM
--(SELECT id,issn_printed as issn FROM ${stats_db_name}.datasource
--WHERE issn_printed IS NOT NULL
--UNION ALL
--SELECT id, issn_online as issn FROM ${stats_db_name}.datasource
--WHERE issn_online IS NOT NULL or id like '%doajarticles%') as issn
--WHERE LENGTH(issn) > 7),
--alljournals AS(select issn, issn_l from stats_ext.alljournals
--where journal_is_in_doaj=true or journal_is_oa=true)
--SELECT DISTINCT pd.id, coalesce(is_gold, 0) as is_gold
--FROM ${stats_db_name}.publication_datasources pd
--left outer join (
--select pd.id, 1 as is_gold FROM ${stats_db_name}.publication_datasources pd
--JOIN issn on issn.id=pd.datasource
--JOIN gold_oa on issn.issn = gold_oa.issn
--join alljournals on issn.issn=alljournals.issn
--left outer join ${stats_db_name}.result_instance ri on ri.id=pd.id
--and ri.accessright!='Closed Access' and ri.accessright_uw='gold') tmp
--on pd.id=tmp.id;
create table if not exists ${stats_db_name}.indi_pub_gold_oa stored as parquet as
with gold_oa as (
select distinct issn from (
SELECT issn_l as issn from stats_ext.issn_gold_oa_dataset_v5
UNION ALL
SELECT issn as issn from stats_ext.issn_gold_oa_dataset_v5
UNION ALL
select issn from stats_ext.alljournals where journal_is_in_doaj=true or journal_is_oa=true
UNION ALL
select issn_l as issn from stats_ext.alljournals where journal_is_in_doaj=true or journal_is_oa=true) foo),
dd as (
select distinct * from (
select id, issn_printed as issn from ${stats_db_name}.datasource d where d.id like '%doajarticles%'
UNION ALL
select id, issn_online as issn from ${stats_db_name}.datasource d where d.id like '%doajarticles%'
UNION ALL
select id, issn_printed as issn from ${stats_db_name}.datasource d join gold_oa on gold_oa.issn=d.issn_printed
UNION ALL
select id, issn_online as issn from ${stats_db_name}.datasource d join gold_oa on gold_oa.issn=d.issn_online) foo
)
SELECT DISTINCT pd.id, coalesce(is_gold, 0) as is_gold
FROM ${stats_db_name}.publication_datasources pd
left outer join (
select pd.id, 1 as is_gold
FROM ${stats_db_name}.publication_datasources pd
join dd on dd.id=pd.datasource
left outer join ${stats_db_name}.result_accessroute ra on ra.id = pd.id where ra.accessroute = 'gold') tmp on tmp.id=pd.id; /*EOS*/
with gold_oa as (
select distinct issn from (
SELECT issn_l as issn from stats_ext.issn_gold_oa_dataset_v5
UNION ALL
SELECT issn as issn from stats_ext.issn_gold_oa_dataset_v5
UNION ALL
select issn from stats_ext.alljournals where journal_is_in_doaj=true or journal_is_oa=true
UNION ALL
select issn_l as issn from stats_ext.alljournals where journal_is_in_doaj=true or journal_is_oa=true) foo),
dd as (
select distinct * from (
select id, issn_printed as issn from ${stats_db_name}.datasource d where d.id like '%doajarticles%'
UNION ALL
select id, issn_online as issn from ${stats_db_name}.datasource d where d.id like '%doajarticles%'
UNION ALL
select id, issn_printed as issn from ${stats_db_name}.datasource d join gold_oa on gold_oa.issn=d.issn_printed
UNION ALL
select id, issn_online as issn from ${stats_db_name}.datasource d join gold_oa on gold_oa.issn=d.issn_online) foo
)
SELECT DISTINCT pd.id, coalesce(is_gold, 0) as is_gold
FROM ${stats_db_name}.publication_datasources pd
left outer join (
select pd.id, 1 as is_gold
FROM ${stats_db_name}.publication_datasources pd
join dd on dd.id=pd.datasource
left outer join ${stats_db_name}.result_accessroute ra on ra.id = pd.id where ra.accessroute = 'gold') tmp on tmp.id=pd.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_hybrid_oa_with_cc purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_pub_hybrid_oa_with_cc stored as parquet as
WITH hybrid_oa AS (
SELECT issn_l, journal_is_in_doaj, journal_is_oa, issn_print as issn
@ -395,121 +260,57 @@ create table if not exists ${stats_db_name}.indi_pub_hybrid_oa_with_cc stored as
FROM STATS_EXT.plan_s_jn
WHERE issn_online != "" and (journal_is_in_doaj = FALSE OR journal_is_oa = FALSE)),
issn AS (
SELECT *
FROM (
SELECT id, issn_printed as issn
FROM ${stats_db_name}.datasource
WHERE issn_printed IS NOT NULL
UNION ALL
SELECT id,issn_online as issn
FROM ${stats_db_name}.datasource
WHERE issn_online IS NOT NULL ) as issn
WHERE LENGTH(issn) > 7)
SELECT DISTINCT pd.id, coalesce(is_hybrid_oa, 0) as is_hybrid_oa
FROM ${stats_db_name}.publication_datasources pd
LEFT OUTER JOIN (
SELECT pd.id, 1 as is_hybrid_oa from ${stats_db_name}.publication_datasources pd
JOIN ${stats_db_name}.datasource d on d.id=pd.datasource
JOIN issn on issn.id=pd.datasource
JOIN hybrid_oa ON issn.issn = hybrid_oa.issn
JOIN ${stats_db_name}.indi_result_has_cc_licence cc on pd.id=cc.id
JOIN ${stats_db_name}.indi_pub_gold_oa ga on pd.id=ga.id
where cc.has_cc_license=1 and ga.is_gold=0) tmp on pd.id=tmp.id; /*EOS*/
SELECT *
FROM (
SELECT id, issn_printed as issn
FROM ${stats_db_name}.datasource
WHERE issn_printed IS NOT NULL
UNION ALL
SELECT id,issn_online as issn
FROM ${stats_db_name}.datasource
WHERE issn_online IS NOT NULL ) as issn
WHERE LENGTH(issn) > 7)
SELECT DISTINCT pd.id, coalesce(is_hybrid_oa, 0) as is_hybrid_oa
FROM ${stats_db_name}.publication_datasources pd
LEFT OUTER JOIN (
SELECT pd.id, 1 as is_hybrid_oa from ${stats_db_name}.publication_datasources pd
JOIN ${stats_db_name}.datasource d on d.id=pd.datasource
JOIN issn on issn.id=pd.datasource
JOIN hybrid_oa ON issn.issn = hybrid_oa.issn
JOIN ${stats_db_name}.indi_result_has_cc_licence cc on pd.id=cc.id
JOIN ${stats_db_name}.indi_pub_gold_oa ga on pd.id=ga.id where cc.has_cc_license=1 and ga.is_gold=0) tmp on pd.id=tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_hybrid purge; /*EOS*/
--create table if not exists ${stats_db_name}.indi_pub_hybrid stored as parquet as
-- WITH gold_oa AS ( SELECT
-- issn_l,
-- journal_is_in_doaj,
-- journal_is_oa,
-- issn_1 as issn,
-- has_apc
-- FROM
-- STATS_EXT.oa_journals
-- WHERE
-- issn_1 != ""
-- UNION
-- ALL SELECT
-- issn_l,
-- journal_is_in_doaj,
-- journal_is_oa,
-- issn_2 as issn,
-- has_apc
-- FROM
-- STATS_EXT.oa_journals
-- WHERE
-- issn_2 != "" ), issn AS ( SELECT
-- *
-- FROM
--( SELECT
-- id,
-- issn_printed as issn
-- FROM
-- ${stats_db_name}.datasource
-- WHERE
-- issn_printed IS NOT NULL
-- UNION ALL
-- SELECT
-- id,
-- issn_online as issn
-- FROM
-- ${stats_db_name}.datasource
-- WHERE
-- issn_online IS NOT NULL or id like '%doajarticles%') as issn
-- WHERE
-- LENGTH(issn) > 7)
--select distinct pd.id, coalesce(is_hybrid, 0) as is_hybrid
--from ${stats_db_name}.publication_datasources pd
-- left outer join (
-- select pd.id, 1 as is_hybrid from ${stats_db_name}.publication_datasources pd
-- join ${stats_db_name}.datasource d on d.id=pd.datasource
-- join issn on issn.id=pd.datasource
-- join gold_oa on issn.issn=gold_oa.issn
-- where (gold_oa.journal_is_in_doaj=false or gold_oa.journal_is_oa=false))tmp
-- on pd.id=tmp.id;
--create table if not exists ${stats_db_name}.indi_pub_hybrid stored as parquet as
--select distinct pd.id,coalesce(is_hybrid,0) is_hybrid from ${stats_db_name}.publication_datasources pd
--left outer join (select pd.id, 1 as is_hybrid from ${stats_db_name}.publication_datasources pd
--join ${stats_db_name}.datasource d on pd.datasource=d.id
--join ${stats_db_name}.result_instance ri on ri.id=pd.id
--join ${stats_db_name}.indi_pub_gold_oa indi_gold on indi_gold.id=pd.id
--join ${stats_db_name}.result_accessroute ra on ra.id=pd.id
--where d.type like '%Journal%' and ri.accessright!='Closed Access' and (ri.accessright_uw!='gold'
--or indi_gold.is_gold=0) and (ra.accessroute='hybrid' or ri.license is not null)) tmp
--on pd.id=tmp.id;
create table if not exists ${stats_db_name}.indi_pub_hybrid stored as parquet as
select distinct pd.id,coalesce(is_hybrid,0) is_hybrid from ${stats_db_name}.publication pd
left outer join (select pd.id, 1 as is_hybrid from ${stats_db_name}.publication pd
join ${stats_db_name}.result_instance ri on ri.id=pd.id
join ${stats_db_name}.indi_pub_gold_oa indi_gold on indi_gold.id=pd.id
join ${stats_db_name}.result_accessroute ra on ra.id=pd.id
join ${stats_db_name}.datasource d on d.id=ri.hostedby
where indi_gold.is_gold=0 and ((d.type like '%Journal%' and ri.accessright!='Closed Access' and ri.accessright!='Restricted' and ri.license is not null) or
ra.accessroute='hybrid'))tmp
on pd.id=tmp.id; /*EOS*/
left outer join (
select pd.id, 1 as is_hybrid from ${stats_db_name}.publication pd
join ${stats_db_name}.result_instance ri on ri.id=pd.id
join ${stats_db_name}.indi_pub_gold_oa indi_gold on indi_gold.id=pd.id
join ${stats_db_name}.result_accessroute ra on ra.id=pd.id
join ${stats_db_name}.datasource d on d.id=ri.hostedby
where indi_gold.is_gold=0 and ((d.type like '%Journal%' and ri.accessright!='Closed Access' and ri.accessright!='Restricted' and ri.license is not null) or ra.accessroute='hybrid')) tmp on pd.id=tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_org_fairness purge; /*EOS*/
create table if not exists ${stats_db_name}.indi_org_fairness stored as parquet as
--return results with PIDs, and rich metadata group by organization
with result_fair as
(select ro.organization organization, count(distinct ro.id) no_result_fair from ${stats_db_name}.result_organization ro
join ${stats_db_name}.result r on r.id=ro.id
with result_fair as (
select ro.organization organization, count(distinct ro.id) no_result_fair
from ${stats_db_name}.result_organization ro
join ${stats_db_name}.result r on r.id=ro.id
--join result_pids rp on r.id=rp.id
where (title is not null) and (publisher is not null) and (abstract=true) and (year is not null) and (authors>0) and cast(year as int)>2003
group by ro.organization),
where (title is not null) and (publisher is not null) and (abstract=true) and (year is not null) and (authors>0) and cast(year as int)>2003
group by ro.organization),
--return all results group by organization
allresults as (select ro.organization, count(distinct ro.id) no_allresults from ${stats_db_name}.result_organization ro
join ${stats_db_name}.result r on r.id=ro.id
where cast(year as int)>2003
group by ro.organization)
allresults as (
select ro.organization, count(distinct ro.id) no_allresults from ${stats_db_name}.result_organization ro
join ${stats_db_name}.result r on r.id=ro.id
where cast(year as int)>2003
group by ro.organization)
--return results_fair/all_results
select allresults.organization, result_fair.no_result_fair/allresults.no_allresults org_fairness
from allresults
join result_fair on result_fair.organization=allresults.organization; /*EOS*/
select allresults.organization, result_fair.no_result_fair/allresults.no_allresults org_fairness
from allresults
join result_fair on result_fair.organization=allresults.organization; /*EOS*/
CREATE TEMPORARY VIEW result_fair as
select ro.organization organization, count(distinct ro.id) no_result_fair
@ -534,8 +335,8 @@ select ar.organization, rf.no_result_fair/ar.no_allresults org_fairness
from allresults ar
join result_fair rf on rf.organization=ar.organization; /*EOS*/
DROP VIEW result_fair; /*EOS*/
DROP VIEW allresults; /*EOS*/
DROP VIEW result_fair;
DROP VIEW allresults;
CREATE TEMPORARY VIEW result_fair as
select year, ro.organization organization, count(distinct ro.id) no_result_fair from ${stats_db_name}.result_organization ro
@ -859,19 +660,6 @@ drop view pub_fos_totals; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_bronze_oa purge; /*EOS*/
--create table if not exists ${stats_db_name}.indi_pub_bronze_oa stored as parquet as
--select distinct p.id, coalesce(is_bronze_oa,0) as is_bronze_oa
--from ${stats_db_name}.publication p
--left outer join
--(select p.id, 1 as is_bronze_oa from ${stats_db_name}.publication p
--join ${stats_db_name}.indi_result_has_cc_licence cc on cc.id=p.id
--join ${stats_db_name}.indi_pub_gold_oa ga on ga.id=p.id
--join ${stats_db_name}.result_instance ri on ri.id=p.id
--join ${stats_db_name}.datasource d on d.id=ri.hostedby
--where cc.has_cc_license=0 and ga.is_gold=0
--and (d.type='Journal' or d.type='Journal Aggregator/Publisher')
--and ri.accessright='Open Access') tmp on tmp.id=p.id;
create table ${stats_db_name}.indi_pub_bronze_oa stored as parquet as
select distinct pd.id,coalesce(is_bronze_oa,0) is_bronze_oa from ${stats_db_name}.publication pd
left outer join (select pd.id, 1 as is_bronze_oa from ${stats_db_name}.publication pd
@ -1189,7 +977,7 @@ union all
select pf.name from stats_ext.insitutions_for_publicly_funded pf
join ${stats_db_name}.project p on p.funder=pf.name
union all
select pf.name from stats_ext.insitutions_for_publicly_funded pf
select op.name from stats_ext.insitutions_for_publicly_funded pf
join org_names_pids op on (op.name=pf.name or op.pid=pf.ror)
and pf.publicly_funded='yes') foo)
select distinct p.id, coalesce(publicly_funded, 0) as publicly_funded
@ -1199,3 +987,53 @@ select distinct ro.id, 1 as publicly_funded from ${stats_db_name}.result_organiz
join ${stats_db_name}.organization o on o.id=ro.organization
join publicly_funded_orgs pfo on o.name=pfo.name) tmp on p.id=tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_pub_green_with_license purge; /*EOS*/
create table ${stats_db_name}.indi_pub_green_with_license stored as parquet as
select distinct p.id, coalesce(green_with_license, 0) as green_with_license
from ${stats_db_name}.publication p
left outer join (
select distinct p.id, 1 as green_with_license from ${stats_db_name}.publication p
join ${stats_db_name}.result_instance ri on ri.id = p.id
join ${stats_db_name}.datasource on datasource.id = ri.hostedby
where ri.license is not null and datasource.type like '%Repository%' and datasource.name!='Other') tmp on p.id= tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.result_country purge; /*EOS*/
create table ${stats_db_name}.result_country stored as parquet as
select distinct ro.id, coalesce(o.country, f.country)
from ${stats_db_name}.result_organization ro
left outer join ${stats_db_name}.organization o on o.id=ro.organization
left outer join ${stats_db_name}.result_projects rp on rp.id=ro.id
left outer join ${stats_db_name}.project p on p.id=rp.project
left outer join ${stats_db_name}.funder f on f.name=p.funder
where coalesce(o.country, f.country) IS NOT NULL;
drop table if exists ${stats_db_name}.indi_result_oa_with_license purge; /*EOS*/
create table ${stats_db_name}.indi_result_oa_with_license stored as parquet as
select distinct r.id, coalesce(oa_with_license,0) as oa_with_license
from ${stats_db_name}.result r
left outer join (select distinct r.id, 1 as oa_with_license from ${stats_db_name}.result r
join ${stats_db_name}.result_licenses rl on rl.id=r.id where r.bestlicence='Open Access') tmp on r.id=tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_result_oa_without_license purge; /*EOS*/
create table ${stats_db_name}.indi_result_oa_without_license stored as parquet as
with without_license as
(select distinct id from ${stats_db_name}.indi_result_oa_with_license
where oa_with_license=0)
select distinct r.id, coalesce(oa_without_license,0) as oa_without_license
from ${stats_db_name}.result r
left outer join (select distinct r.id, 1 as oa_without_license
from ${stats_db_name}.result r
join without_license wl on wl.id=r.id
where r.bestlicence='Open Access') tmp on r.id=tmp.id; /*EOS*/
drop table if exists ${stats_db_name}.indi_result_under_transformative purge; /*EOS*/
create table ${stats_db_name}.indi_result_under_transformative stored as parquet as
with transformative_dois as (
select distinct doi from stats_ext.transformative_facts)
select distinct r.id, coalesce(under_transformative,0) as under_transformative
from ${stats_db_name}.result r
left outer join (
select distinct rp.id, 1 as under_transformative
from ${stats_db_name}.result_pids rp join ${stats_db_name}.result r on r.id=rp.id
join transformative_dois td on td.doi=rp.pid) tmp on r.id=tmp.id; /*EOS*/