Project

General

Profile

« Previous | Next » 

Revision 50431

updated sql queries used to map datasources and organizations to HBase

View differences:

modules/dnet-openaireplus-workflows/branches/dsm/src/main/resources/eu/dnetlib/msro/openaireplus/workflows/hbase/queryDatasources.sql
1
EXPLAIN SELECT
1
SELECT
2 2
	d.id                                                                                                       AS datasourceid,
3 3
	d.id || array_agg(di.pid)                                                                                  AS identities,
4 4
	d.officialname                                                                                             AS officialname,
......
65 65
	d.qualitymanagementkind                                                                                    AS qualitymanagementkind,
66 66
	d.pidsystems                                                                                               AS pidsystems,
67 67
	d.certificates                                                                                             AS certificates,
68
	'{}'::text[]                                                                                               AS policies,
68
	ARRAY[]::text[]                                                                                            AS policies,
69 69
	dc.id                                                                                                      AS collectedfromid,
70 70
	dc.officialname                                                                                            AS collectedfromname,
71 71
	d.typology || '@@@' || d.typology || '@@@dnet:datasource_typologies@@@dnet:datasource_typologies'          AS datasourcetype,
modules/dnet-openaireplus-workflows/branches/dsm/src/main/resources/eu/dnetlib/msro/openaireplus/workflows/hbase/queryDatasourceOrganization.sql
1 1
SELECT
2 2
	dor.datasource                                                          AS datasource,
3 3
	dor.organization                                                        AS organization,
4
	dor.startdate                                                           AS startdate,
5
	dor.enddate                                                             AS enddate,
6
	dor.inferred                                                            AS inferred,
7
	dor.deletedbyinference                                                  AS deletedbyinference,
8
	dor.trust                                                               AS trust,
9
	dor.inferenceprovenance                                                 AS inferenceprovenance,
4
	NULL                                                                    AS startdate,
5
	NULL                                                                    AS enddate,
6
	false                                                                   AS inferred,
7
	false                                                                   AS deletedbyinference,
8
	0.9                                                                     AS trust,
9
	NULL                                                                    AS inferenceprovenance,
10 10

  
11
	doc.code || '@@@' || doc.name || '@@@' || dos.code || '@@@' || dos.name AS semantics,
12
	pac.code || '@@@' || pac.name || '@@@' || pas.code || '@@@' || pas.name AS provenanceaction
11
	'providedBy@@@provided by@@@dnet:datasources_organizations_typologies@@@dnet:datasources_organizations_typologies' AS semantics,
12
	d.provenanceaction || '@@@' || d.provenanceaction || '@@@dnet:provenanceActions@@@dnet:provenanceActions' AS provenanceaction
13 13

  
14
FROM datasource_organization dor
14
FROM dsm_datasource_organization dor
15
	LEFT OUTER JOIN dsm_datasources d ON (dor.datasource = d.id)
15 16

  
16
	LEFT OUTER JOIN datasources d ON (dor.datasource = d.id)
17

  
18
	LEFT OUTER JOIN class pac ON (pac.code = d.provenanceactionclass)
19
	LEFT OUTER JOIN scheme pas ON (pas.code = d.provenanceactionscheme)
20

  
21
	LEFT OUTER JOIN class doc ON (doc.code = dor.semanticclass)
22
	LEFT OUTER JOIN scheme dos ON (dos.code = dor.semanticscheme)
23

  
24
--  These clauses have been commented to allow the import of all the datasource on the index 
25
-- WHERE d.datasourceclass != 'entityregistry'
26
--	AND d.openairecompatibilityclass IS NOT null
27
--	AND d.openairecompatibilityclass != 'UNKNOWN'
28
--	AND d.openairecompatibilityclass != 'notCompatible'
modules/dnet-openaireplus-workflows/branches/dsm/src/main/resources/eu/dnetlib/msro/openaireplus/workflows/hbase/queryDatasourcesForSize.sql
1 1
SELECT count(*)
2
FROM datasources;
2
FROM dsm_datasources;

Also available in: Unified diff