diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step2.sql b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step2.sql index 75b24b189..bb0d0ac6c 100644 --- a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step2.sql +++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step2.sql @@ -90,27 +90,8 @@ FROM ${openaire_db_name}.publication p where p.datainfo.deletedbyinference = false; CREATE TABLE ${stats_db_name}.publication_citations AS -SELECT substr(p.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS result +SELECT substr(p.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS cites FROM ${openaire_db_name}.publication p lateral view explode(p.extrainfo) citations AS citation WHERE xpath_string(citation.value, "//citation/id[@type='openaire']/@value") != "" - and p.datainfo.deletedbyinference = false; - --- ANALYZE TABLE ${stats_db_name}.publication_tmp COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.publication_tmp COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.publication_classifications COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.publication_classifications COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.publication_concepts COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.publication_concepts COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.publication_datasources COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.publication_datasources COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.publication_languages COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.publication_languages COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.publication_oids COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.publication_oids COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.publication_pids COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.publication_pids COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.publication_topics COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.publication_topics COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.publication_citations COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.publication_citations COMPUTE STATISTICS FOR COLUMNS; \ No newline at end of file + and p.datainfo.deletedbyinference = false; \ No newline at end of file diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step3.sql b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step3.sql index 540cc03a5..953eaad6a 100644 --- a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step3.sql +++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step3.sql @@ -41,7 +41,7 @@ FROM ${openaire_db_name}.dataset d WHERE d.datainfo.deletedbyinference = FALSE; CREATE TABLE ${stats_db_name}.dataset_citations AS -SELECT substr(d.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS result +SELECT substr(d.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS cites FROM ${openaire_db_name}.dataset d LATERAL VIEW explode(d.extrainfo) citations AS citation WHERE xpath_string(citation.value, "//citation/id[@type='openaire']/@value") != "" @@ -95,21 +95,4 @@ CREATE TABLE ${stats_db_name}.dataset_topics AS SELECT substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS type, subjects.subject.value AS topic FROM ${openaire_db_name}.dataset p LATERAL VIEW explode(p.subject) subjects AS subject -where p.datainfo.deletedbyinference = false; --- --- ANALYZE TABLE ${stats_db_name}.dataset_tmp COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.dataset_tmp COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.dataset_classifications COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.dataset_classifications COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.dataset_concepts COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.dataset_concepts COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.dataset_datasources COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.dataset_datasources COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.dataset_languages COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.dataset_languages COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.dataset_oids COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.dataset_oids COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.dataset_pids COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.dataset_pids COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.dataset_topics COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.dataset_topics COMPUTE STATISTICS FOR COLUMNS; \ No newline at end of file +where p.datainfo.deletedbyinference = false; \ No newline at end of file diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step4.sql b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step4.sql index 54345e074..0210dc8cb 100644 --- a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step4.sql +++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step4.sql @@ -41,7 +41,7 @@ from ${openaire_db_name}.software s where s.datainfo.deletedbyinference = false; CREATE TABLE ${stats_db_name}.software_citations AS -SELECT substr(s.id, 4) as id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS RESULT +SELECT substr(s.id, 4) as id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS cites FROM ${openaire_db_name}.software s LATERAL VIEW explode(s.extrainfo) citations as citation where xpath_string(citation.value, "//citation/id[@type='openaire']/@value") != "" @@ -95,21 +95,4 @@ CREATE TABLE ${stats_db_name}.software_topics AS SELECT substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS type, subjects.subject.value AS topic FROM ${openaire_db_name}.software p LATERAL VIEW explode(p.subject) subjects AS subject -where p.datainfo.deletedbyinference = false; --- --- ANALYZE TABLE ${stats_db_name}.software_tmp COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.software_tmp COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.software_classifications COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.software_classifications COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.software_concepts COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.software_concepts COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.software_datasources COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.software_datasources COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.software_languages COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.software_languages COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.software_oids COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.software_oids COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.software_pids COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.software_pids COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.software_topics COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.software_topics COMPUTE STATISTICS FOR COLUMNS; \ No newline at end of file +where p.datainfo.deletedbyinference = false; \ No newline at end of file diff --git a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step5.sql b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step5.sql index 36ad5d92a..f7b302186 100644 --- a/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step5.sql +++ b/dhp-workflows/dhp-stats-update/src/main/resources/eu/dnetlib/dhp/oa/graph/stats/oozie_app/scripts/step5.sql @@ -41,7 +41,7 @@ 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 +SELECT substr(o.id, 4) AS id, xpath_string(citation.value, "//citation/id[@type='openaire']/@value") AS cites 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; @@ -86,21 +86,4 @@ where p.datainfo.deletedbyinference = false; CREATE TABLE ${stats_db_name}.otherresearchproduct_topics AS SELECT substr(p.id, 4) AS id, subjects.subject.qualifier.classname AS type, subjects.subject.value AS topic FROM ${openaire_db_name}.otherresearchproduct p LATERAL VIEW explode(p.subject) subjects AS subject -where p.datainfo.deletedbyinference = false; - --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_tmp COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_tmp COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_classifications COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_classifications COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_concepts COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_concepts COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_datasources COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_datasources COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_languages COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_languages COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_oids COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_oids COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_pids COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_pids COMPUTE STATISTICS FOR COLUMNS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_topics COMPUTE STATISTICS; --- ANALYZE TABLE ${stats_db_name}.otherresearchproduct_topics COMPUTE STATISTICS FOR COLUMNS; \ No newline at end of file +where p.datainfo.deletedbyinference = false; \ No newline at end of file