miriam.baglioni
/
dnet-hadoop
forked from D-Net/dnet-hadoop
2
0
Fork 0
Code Issues 2 Pull Requests Releases Wiki Activity

added sprint 6 indicators to monitor db

This commit is contained in:
Antonis Lempesis 2022-02-18 17:11:23 +02:00 committed by dimitrispie
parent 58c59f46eb
commit 87c91f70a2
2 changed files with 187 additions and 166 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

144
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step16-createIndicatorsTables.sql
View File

@ -31,17 +31,6 @@ join datasource d on d.id = ri.collectedfrom
where pidtype='Digital Object Identifier' and d.name ='Crossref') tmp
on tmp.id=p.id;
create table indi_pub_gold_oa stored as parquet as
select distinct p.id, coalesce(gold_oa, 0) as gold_oa
from publication p
left outer join (
select p.id, 1 as gold_oa
from publication p
join result_instance ri on ri.id = p.id
join datasource on datasource.id = ri.hostedby
where datasource.id like '%doajarticles%') tmp
on p.id= tmp.id;
---- Sprint 2 ----
create table 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
@ -144,16 +133,6 @@ join stats_ext.plan_s_jn ps where (ps.issn_print=d.issn_printed and ps.issn_onli
and (ps.journal_is_in_doaj=false and ps.journal_is_oa=false)) tmp
on pd.id=tmp.id;
create table indi_is_gold_oa stored as parquet as
(select distinct pd.id, coalesce(gold_oa, 0) as gold_oa
from publication_datasources pd
left outer join (
select pd.id, 1 as gold_oa from publication_datasources pd
join datasource d on d.id=pd.datasource
join stats_ext.plan_s_jn ps on (ps.issn_print=d.issn_printed or ps.issn_online=d.issn_online)
where ps.journal_is_in_doaj is true or ps.journal_is_oa is true) tmp
on pd.id=tmp.id);
create table indi_pub_in_transformative stored as parquet as
select distinct pd.id, coalesce(is_transformative, 0) as is_transformative
from publication pd
@ -180,15 +159,17 @@ create table indi_result_no_of_copies stored as parquet as
select id, count(id) as number_of_copies from result_instance group by id;
---- Sprint 6 ----
create table indi_pub_gold_oa_new stored as parquet as
WITH gold_oa AS (SELECT issn_l, journal_is_in_doaj,journal_is_oa, issn_1 as issn
create table 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
issn AS (
SELECT * FROM
(SELECT id, issn_printed as issn
FROM datasource WHERE issn_printed IS NOT NULL
UNION
@ -197,23 +178,26 @@ FROM datasource WHERE issn_online IS NOT NULL) as issn
WHERE LENGTH(issn) > 7)
SELECT DISTINCT pd.id, coalesce(is_gold, 0) as is_gold
FROM publication_datasources pd
LEFT OUTER JOIN
(SELECT pd.id, 1 as is_gold FROM publication_datasources pd
LEFT OUTER JOIN (
SELECT pd.id, 1 as is_gold FROM 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;
JOIN gold_oa on issn.issn = gold_oa.issn) tmp ON pd.id=tmp.id;
create table indi_datasets_gold_oa_new stored as parquet as
WITH gold_oa AS (SELECT issn_l, journal_is_in_doaj, journal_is_oa, issn_1 as issn
create table indi_datasets_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 datasource WHERE issn_printed IS NOT NULL
issn AS (
SELECT *
FROM (
SELECT id,issn_printed as issn
FROM datasource
WHERE issn_printed IS NOT NULL
UNION
SELECT id, issn_online as issn
FROM datasource
@ -221,23 +205,26 @@ WHERE issn_online IS NOT NULL ) as issn
WHERE LENGTH(issn) > 7)
SELECT DISTINCT pd.id, coalesce(is_gold, 0) as is_gold
FROM dataset_datasources pd
LEFT OUTER JOIN
(SELECT pd.id, 1 as is_gold FROM dataset_datasources pd
LEFT OUTER JOIN (
SELECT pd.id, 1 as is_gold FROM dataset_datasources pd
JOIN issn on issn.id=pd.datasource
JOIN gold_oa on issn.issn = gold_oa.issn) tmp
ON pd.id=tmp.id;
JOIN gold_oa on issn.issn = gold_oa.issn) tmp ON pd.id=tmp.id;
create table indi_software_gold_oa_new stored as parquet as
WITH gold_oa AS (SELECT issn_l, journal_is_in_doaj, journal_is_oa, issn_1 as issn
create table indi_software_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 datasource WHERE issn_printed IS NOT NULL
issn AS (
SELECT *
FROM (
SELECT id,issn_printed as issn
FROM datasource
WHERE issn_printed IS NOT NULL
UNION
SELECT id, issn_online as issn
FROM datasource
@ -245,22 +232,22 @@ WHERE issn_online IS NOT NULL ) as issn
WHERE LENGTH(issn) > 7)
SELECT DISTINCT pd.id, coalesce(is_gold, 0) as is_gold
FROM software_datasources pd
LEFT OUTER JOIN
(SELECT pd.id, 1 as is_gold FROM software_datasources pd
LEFT OUTER JOIN (
SELECT pd.id, 1 as is_gold FROM software_datasources pd
JOIN issn on issn.id=pd.datasource
JOIN gold_oa on issn.issn = gold_oa.issn) tmp
ON pd.id=tmp.id;
JOIN gold_oa on issn.issn = gold_oa.issn) tmp ON pd.id=tmp.id;
create table indi_org_findable stored as parquet as
with result_with_pid as
(select ro.organization organization, count(distinct rp.id) no_result_with_pid from result_organization ro
with result_with_pid as (
select ro.organization organization, count(distinct rp.id) no_result_with_pid from result_organization ro
join result_pids rp on rp.id=ro.id
group by ro.organization),
result_has_abstract as
(select ro.organization organization, count(distinct rp.id) no_result_with_abstract from result_organization ro
result_has_abstract as (
select ro.organization organization, count(distinct rp.id) no_result_with_abstract from result_organization ro
join result rp on rp.id=ro.id where rp.abstract=true
group by ro.organization),
allresults as (select organization, count(distinct id) no_allresults from result_organization
allresults as (
select organization, count(distinct id) no_allresults from result_organization
group by organization),
result_with_pid_share as (
select allresults.organization, result_with_pid.no_result_with_pid/allresults.no_allresults pid_share
@ -274,9 +261,7 @@ select allresults.organization, coalesce((pid_share+abstract_share)/2,pid_share)
from allresults
join result_with_pid_share on result_with_pid_share.organization=allresults.organization
left outer join (
select organization, abstract_share from result_with_abstract_share
)tmp
on tmp.organization=allresults.organization;
select organization, abstract_share from result_with_abstract_share) tmp on tmp.organization=allresults.organization;
create table indi_org_openess stored as parquet as
WITH datasets_oa as (
@ -320,24 +305,27 @@ allsoftwaresshare as (
select software_oa.organization, software_oa.no_oasoftware/allsoftware.no_allsoftware s
from allsoftware
join software_oa on allsoftware.organization=software_oa.organization)
select allpubsshare.organization, coalesce((c+p+s)/3, p) org_openess FROM allpubsshare
left outer join (select organization,c from
alldatasetssshare) tmp
on tmp.organization=allpubsshare.organization
left outer join (select organization,s from
allsoftwaresshare) tmp1
on tmp1.organization=allpubsshare.organization;
select allpubsshare.organization, coalesce((c+p+s)/3, p) org_openess
FROM allpubsshare
left outer join (
select organization,c from
alldatasetssshare) tmp on tmp.organization=allpubsshare.organization
left outer join (
select organization,s from allsoftwaresshare) tmp1 on tmp1.organization=allpubsshare.organization;
create table 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
WITH hybrid_oa AS (
SELECT issn_l, journal_is_in_doaj, journal_is_oa, issn_print as issn
FROM stats_ext.plan_s_jn
WHERE issn_print != ""
UNION ALL
SELECT issn_l, journal_is_in_doaj, journal_is_oa, issn_online as issn
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
issn AS (
SELECT *
FROM (
SELECT id, issn_printed as issn
FROM datasource
WHERE issn_printed IS NOT NULL
UNION
@ -353,7 +341,31 @@ JOIN datasource d on d.id=pd.datasource
JOIN issn on issn.id=pd.datasource
JOIN hybrid_oa ON issn.issn = hybrid_oa.issn
JOIN indi_result_has_cc_licence cc on pd.id=cc.id
where cc.has_cc_license=1) tmp
on pd.id=tmp.id;
where cc.has_cc_license=1) tmp on pd.id=tmp.id;
create table indi_pub_downloads stored as parquet as
SELECT result_id, sum(downloads) no_dowloads from openaire_prod_usage_stats.usage_stats
join publication on result_id=id
where downloads>0
GROUP BY result_id
order by no_dowloads desc;
create table indi_pub_downloads_datasource stored as parquet as
SELECT result_id, repository_id, sum(downloads) no_dowloads from openaire_prod_usage_stats.usage_stats
join publication on result_id=id
where downloads>0
GROUP BY result_id, repository_id
order by result_id;
create table indi_pub_downloads_year stored as parquet as
SELECT result_id, substring(us.`date`, 1,4) as `year`, sum(downloads) no_dowloads from openaire_prod_usage_stats.usage_stats us
join publication on result_id=id where downloads>0
GROUP BY result_id, `year`
order by `year` asc;
create table indi_pub_downloads_datasource_year stored as parquet as
SELECT result_id, substring(us.`date`, 1,4) as `year`, repository_id, sum(downloads) no_dowloads from openaire_prod_usage_stats.usage_stats us
join publication on result_id=id
where downloads>0
GROUP BY result_id, repository_id, `year`
order by `year` asc, result_id;

9
dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step20-createMonitorDB.sql
View File

@ -123,6 +123,10 @@ create table TARGET.indi_pub_doi_from_crossref stored as parquet as select * fro
compute stats TARGET.indi_pub_doi_from_crossref;
create table TARGET.indi_pub_gold_oa stored as parquet as select * from SOURCE.indi_pub_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_pub_gold_oa;
create table TARGET.indi_datasets_gold_oa stored as parquet as select * from SOURCE.indi_datasets_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_datasets_gold_oa;
create table TARGET.indi_software_gold_oa stored as parquet as select * from SOURCE.indi_software_gold_oa orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_software_gold_oa;
create table TARGET.indi_pub_has_abstract stored as parquet as select * from SOURCE.indi_pub_has_abstract orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_pub_has_abstract;
create table TARGET.indi_result_has_cc_licence stored as parquet as select * from SOURCE.indi_result_has_cc_licence orig where exists (select 1 from TARGET.result r where r.id=orig.id);
@ -148,6 +152,11 @@ compute stats TARGET.indi_pub_closed_other_open;
create table TARGET.indi_result_no_of_copies stored as parquet as select * from SOURCE.indi_result_no_of_copies orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_result_no_of_copies;
create view TARGET.indi_org_findable as select * from SOURCE.indi_org_findable;
create view TARGET.indi_org_openess as select * from SOURCE.indi_org_openess;
create table TARGET.indi_pub_hybrid_oa_with_cc stored as parquet as select * from SOURCE.indi_pub_hybrid_oa_with_cc orig where exists (select 1 from TARGET.result r where r.id=orig.id);
compute stats TARGET.indi_pub_hybrid_oa_with_cc;
--- Usage statistics
create table TARGET.usage_stats stored as parquet as select * from SOURCE.usage_stats orig where exists (select 1 from TARGET.result r where r.id=orig.result_id);